arXiv Paper Daily: Tue, 31 Jan 2017

Neural and Evolutionary Computing

PathNet: Evolution Channels Gradient Descent in Super Neural Networks

Chrisantha Fernando , Dylan Banarse , Charles Blundell , Yori Zwols , David Ha , Andrei A. Rusu , Alexander Pritzel , Daan Wierstra Subjects : Neural and Evolutionary Computing (cs.NE) ; Learning (cs.LG)

For artificial general intelligence (AGI) it would be efficient if multiple

users trained the same giant neural network, permitting parameter reuse,

without catastrophic forgetting. PathNet is a first step in this direction. It

is a neural network algorithm that uses agents embedded in the neural network

whose task is to discover which parts of the network to re-use for new tasks.

Agents are pathways (views) through the network which determine the subset of

parameters that are used and updated by the forwards and backwards passes of

the backpropogation algorithm. During learning, a tournament selection genetic

algorithm is used to select pathways through the neural network for replication

and mutation. Pathway fitness is the performance of that pathway measured

according to a cost function. We demonstrate successful transfer learning;

fixing the parameters along a path learned on task A and re-evolving a new

population of paths for task B, allows task B to be learned faster than it

could be learned from scratch or after fine-tuning. Paths evolved on task B

re-use parts of the optimal path evolved on task A. Positive transfer was

demonstrated for binary MNIST, CIFAR, and SVHN supervised learning

classification tasks, and a set of Atari and Labyrinth reinforcement learning

tasks, suggesting PathNets have general applicability for neural network

training. Finally, PathNet also significantly improves the robustness to

hyperparameter choices of a parallel asynchronous reinforcement learning

algorithm (A3C).

Memory Augmented Neural Networks with Wormhole Connections

Caglar Gulcehre , Sarath Chandar , Yoshua Bengio Subjects : Learning (cs.LG) ; Neural and Evolutionary Computing (cs.NE); Machine Learning (stat.ML)

Recent empirical results on long-term dependency tasks have shown that neural

networks augmented with an external memory can learn the long-term dependency

tasks more easily and achieve better generalization than vanilla recurrent

neural networks (RNN). We suggest that memory augmented neural networks can

reduce the effects of vanishing gradients by creating shortcut (or wormhole)

connections. Based on this observation, we propose a novel memory augmented

neural network model called TARDIS (Temporal Automatic Relation Discovery in

Sequences). The controller of TARDIS can store a selective set of embeddings of

its own previous hidden states into an external memory and revisit them as and

when needed. For TARDIS, memory acts as a storage for wormhole connections to

the past to propagate the gradients more effectively and it helps to learn the

temporal dependencies. The memory structure of TARDIS has similarities to both

Neural Turing Machines (NTM) and Dynamic Neural Turing Machines (D-NTM), but

both read and write operations of TARDIS are simpler and more efficient. We use

discrete addressing for read/write operations which helps to substantially to

reduce the vanishing gradient problem with very long sequences. Read and write

operations in TARDIS are tied with a heuristic once the memory becomes full,

and this makes the learning problem simpler when compared to NTM or D-NTM type

of architectures. We provide a detailed analysis on the gradient propagation in

general for MANNs. We evaluate our models on different long-term dependency

tasks and report competitive results in all of them.

Source localization in an ocean waveguide using supervised machine learning

Source localization is solved as a classification problem by training a

feed-forward neural network (FNN) on ocean acoustic data. The pressure received

by a vertical linear array is preprocessed by constructing a normalized sample

covariance matrix (SCM), which is used as input for the FNN. Each neuron of the

output layer represents a discrete source range. FNN is a data-driven method

that learns features directly from observed acoustic data, unlike model-based

localization methods such as matched-field processing that require accurate

sound propagation modeling. The FNN achieves a good performance (the mean

absolute percentage error below 10/%) for predicting source ranges for vertical

array data from the Noise09 experiment. The effects of varying the parameters

of the method, such as number of hidden neurons and layers, number of output

neurons and number of snapshots in each input sample are discussed.

Detection, Segmentation and Recognition of Face and its Features Using Neural Network

Face detection and recognition has been prevalent with research scholars and

diverse approaches have been incorporated till date to serve purpose. The

rampant advent of biometric analysis systems, which may be full body scanners,

or iris detection and recognition systems and the finger print recognition

systems, and surveillance systems deployed for safety and security purposes

have contributed to inclination towards same. Advances has been made with

frontal view, lateral view of the face or using facial expressions such as

anger, happiness and gloominess, still images and video image to be used for

detection and recognition. This led to newer methods for face detection and

recognition to be introduced in achieving accurate results and economically

feasible and extremely secure. Techniques such as Principal Component analysis

(PCA), Independent component analysis (ICA), Linear Discriminant Analysis

(LDA), have been the predominant ones to be used. But with improvements needed

in the previous approaches Neural Networks based recognition was like boon to

the industry. It not only enhanced the recognition but also the efficiency of

the process. Choosing Backpropagation as the learning method was clearly out of

its efficiency to recognize nonlinear faces with an acceptance ratio of more

than 90% and execution time of only few seconds.

Computer Vision and Pattern Recognition

Document Decomposition of Bangla Printed Text

Today all kind of information is getting digitized and along with all this

digitization, the huge archive of various kinds of documents is being digitized

too. We know that, Optical Character Recognition is the method through which,

newspapers and other paper documents convert into digital resources. But, it is

a fact that this method works on texts only. As a result, if we try to process

any document which contains non-textual zones, then we will get garbage texts

as output. That is why; in order to digitize documents properly they should be

prepossessed carefully. And while preprocessing, segmenting document in

different regions according to the category properly is most important. But,

the Optical Character Recognition processes available for Bangla language have

no such algorithm that can categorize a newspaper/book page fully. So we worked

to decompose a document into its several parts like headlines, sub headlines,

columns, images etc. And if the input is skewed and rotated, then the input was

also deskewed and de-rotated. To decompose any Bangla document we found out the

edges of the input image. Then we find out the horizontal and vertical area of

every pixel where it lies in. Later on the input image was cut according to

these areas. Then we pick each and every sub image and found out their

height-width ratio, line height. Then according to these values the sub images

were categorized. To deskew the image we found out the skew angle and de skewed

the image according to this angle. To de-rotate the image we used the line

height, matra line, pixel ratio of matra line.

Self-Adaptation of Activity Recognition Systems to New Sensors

Traditional activity recognition systems work on the basis of training,

taking a fixed set of sensors into account. In this article, we focus on the

question how pattern recognition can leverage new information sources without

any, or with minimal user input. Thus, we present an approach for opportunistic

activity recognition, where ubiquitous sensors lead to dynamically changing

input spaces. Our method is a variation of well-established principles of

machine learning, relying on unsupervised clustering to discover structure in

data and inferring cluster labels from a small number of labeled dates in a

semi-supervised manner. Elaborating the challenges, evaluations of over 3000

sensor combinations from three multi-user experiments are presented in detail

and show the potential benefit of our approach.

A Survey on Structure from Motion

The structure from motion (SfM) problem in computer vision is the problem of

recovering the (3)D structure of a stationary scene from a set of projective

measurements, represented as a collection of (2)D images, via estimation of

motion of the cameras corresponding to these images. In essence, SfM involves

the three main stages of (1) extraction of features in images (e.g., points of

interest, lines, etc.) and matching of these features between images, (2)

camera motion estimation (e.g., using relative pairwise camera poses estimated

from the extracted features), (3) recovery of the (3)D structure using the

estimated motion and features (e.g., by minimizing the so-called reprojection

error). This survey mainly focuses on the relatively recent developments in the

literature pertaining to stages (2) and (3). More specifically, after touching

upon the early factorization-based techniques for motion and structure

estimation, we provide a detailed account of some of the recent camera location

estimation methods in the literature, which precedes the discussion of notable

techniques for (3)D structure recovery. We also cover the basics of the

simultaneous localization and mapping (SLAM) problem, which can be considered

to be a specific case of the SfM problem. Additionally, a review of the

fundamentals of feature extraction and matching (i.e., stage (1) above),

various recent methods for handling ambiguities in (3)D scenes, SfM techniques

involving relatively uncommon camera models and image features, and popular

sources of data and SfM software is included in our survey.

CNN as Guided Multi-layer RECOS Transform

C.-C. Jay Kuo Subjects : Computer Vision and Pattern Recognition (cs.CV)

There is a resurging interest in developing a neural-network-based solution

to the supervised machine learning problem. The convolutional neural network

(CNN), which is also known as the feedforward neural network and the

multi-layer perceptron (MLP), will be studied in this note. To begin with, we

introduce a RECOS transform as a basic building block of CNNs. The “RECOS” is

an acronym for “REctified-COrrelations on a Sphere”. It consists of two main

concepts: 1) data clustering on a sphere and 2) rectification. Afterwards, we

interpret a CNN as a network that implements the guided multi-layer RECOS

transform with three highlights. First, we compare the traditional single-layer

and modern multi-layer signal analysis approaches, point out key ingredients

that enable the multi-layer approach, and provide a full explanation to the

operating principle of CNNs. Second, we discuss how guidance is provided by

labels through backpropagation in the training. Third, we show that a trained

network can be greatly simplified in the testing stage demanding only one-bit

representation for both filter weights and inputs.

Scalable Nearest Neighbor Search based on kNN Graph

Nearest neighbor search is known as a challenging issue that has been studied

for several decades. Recently, this issue becomes more and more imminent in

viewing that the big data problem arises from various fields. In this paper, a

scalable solution based on hill-climbing strategy with the support of k-nearest

neighbor graph (kNN) is presented. Two major issues have been considered in the

paper. Firstly, an efficient kNN graph construction method based on two means

tree is presented. For the nearest neighbor search, an enhanced hill-climbing

procedure is proposed, which sees considerable performance boost over original

procedure. Furthermore, with the support of inverted indexing derived from

residue vector quantization, our method achieves close to 100% recall with high

speed efficiency in two state-of-the-art evaluation benchmarks. In addition, a

comparative study on both the compressional and traditional nearest neighbor

search methods is presented. We show that our method achieves the best

trade-off between search quality, efficiency and memory complexity.

Re-ranking Person Re-identification with k-reciprocal Encoding

Zhun Zhong , Liang Zheng , Donglin Cao , Shaozi Li Subjects : Computer Vision and Pattern Recognition (cs.CV)

When considering person re-identification (re-ID) as a retrieval process,

re-ranking is a critical step to improve its accuracy. Yet in the re-ID

community, limited effort has been devoted to re-ranking, especially those

fully automatic, unsupervised solutions. In this paper, we propose a

k-reciprocal encoding method to re-rank the re-ID results. Our hypothesis is

that if a gallery image is similar to the probe in the k-reciprocal nearest

neighbors, it is more likely to be a true match. Specifically, given an image,

a k-reciprocal feature is calculated by encoding its k-reciprocal nearest

neighbors into a single vector, which is used for re-ranking under the Jaccard

distance. The final distance is computed as the combination of the original

distance and the Jaccard distance. Our re-ranking method does not require any

human interaction or any labeled data, so it is applicable to large-scale

datasets. Experiments on the large-scale Market-1501, CUHK03, MARS, and PRW

datasets confirm the effectiveness of our method.

Faceness-Net: Face Detection through Deep Facial Part Responses

We propose a deep convolutional neural network (CNN) for face detection

leveraging on facial attributes based supervision. We observe a phenomenon that

part detectors emerge within CNN trained to classify attributes from uncropped

face images, without any explicit part supervision. The observation motivates a

new method for finding faces through scoring facial parts responses by their

spatial structure and arrangement. The scoring mechanism is data-driven, and

carefully formulated considering challenging cases where faces are only

partially visible. This consideration allows our network to detect faces under

severe occlusion and unconstrained pose variations. Our method achieves

promising performance on popular benchmarks including FDDB, PASCAL Faces, AFW,

and WIDER FACE.

The HASYv2 dataset

Martin Thoma Subjects : Computer Vision and Pattern Recognition (cs.CV)

This paper describes the HASYv2 dataset. HASY is a publicly available, free

of charge dataset of single symbols similar to MNIST. It contains 168233

instances of 369 classes. HASY contains two challenges: A classification

challenge with 10 pre-defined folds for 10-fold cross-validation and a

verification challenge.

MSCM-LiFe: Multi-scale cross modal linear feature for horizon detection in maritime images

This paper proposes a new method for horizon detection called the multi-scale

cross modal linear feature. This method integrates three different concepts

related to the presence of horizon in maritime images to increase the accuracy

of horizon detection. Specifically it uses the persistence of horizon in

multi-scale median filtering, and its detection as a linear feature commonly

detected by two different methods, namely the Hough transform of edgemap and

the intensity gradient. We demonstrate the performance of the method over 13

videos comprising of more than 3000 frames and show that the proposed method

detects horizon with small error in most of the cases, outperforming three

state-of-the-art methods.

VINet: Visual-Inertial Odometry as a Sequence-to-Sequence Learning Problem

In this paper we present an on-manifold sequence-to-sequence learning

approach to motion estimation using visual and inertial sensors. It is to the

best of our knowledge the first end-to-end trainable method for visual-inertial

odometry which performs fusion of the data at an intermediate

feature-representation level. Our method has numerous advantages over

traditional approaches. Specifically, it eliminates the need for tedious manual

synchronization of the camera and IMU as well as eliminating the need for

manual calibration between the IMU and camera. A further advantage is that our

model naturally and elegantly incorporates domain specific information which

significantly mitigates drift. We show that our approach is competitive with

state-of-the-art traditional methods when accurate calibration data is

available and can be trained to outperform them in the presence of calibration

and synchronization errors.

Feature base fusion for splicing forgery detection based on neuro fuzzy

Habib Ghaffari Hadigheh , Ghazali bin sulong Subjects : Computer Vision and Pattern Recognition (cs.CV) ; Artificial Intelligence (cs.AI); Learning (cs.LG)

Most of researches on image forensics have been mainly focused on detection

of artifacts introduced by a single processing tool. They lead in the

development of many specialized algorithms looking for one or more particular

footprints under specific settings. Naturally, the performance of such

algorithms are not perfect, and accordingly the provided output might be noisy,

inaccurate and only partially correct. Furthermore, a forged image in practical

scenarios is often the result of utilizing several tools available by

image-processing software systems. Therefore, reliable tamper detection

requires developing more poweful tools to deal with various tempering

scenarios. Fusion of forgery detection tools based on Fuzzy Inference System

has been used before for addressing this problem. Adjusting the membership

functions and defining proper fuzzy rules for attaining to better results are

time-consuming processes. This can be accounted as main disadvantage of fuzzy

inference systems. In this paper, a Neuro-Fuzzy inference system for fusion of

forgery detection tools is developed. The neural network characteristic of

these systems provides appropriate tool for automatically adjusting the

membership functions. Moreover, initial fuzzy inference system is generated

based on fuzzy clustering techniques. The proposed framework is implemented and

validated on a benchmark image splicing data set in which three forgery

detection tools are fused based on adaptive Neuro-Fuzzy inference system. The

outcome of the proposed method reveals that applying Neuro Fuzzy inference

systems could be a better approach for fusion of forgery detection tools.

Supervised Multilayer Sparse Coding Networks for Image Classification

Xiaoxia Sun , Nasser M. Nasrabadi , Trac D. Tran Subjects : Computer Vision and Pattern Recognition (cs.CV)

In this paper, we propose a novel multilayer sparse coding network capable of

efficiently adapting its own regularization parameters to a given dataset. The

network is trained end-to-end with a supervised task-driven learning algorithm

via error backpropagation. During training, the network learns both the

dictionaries and the regularization parameters of each sparse coding layer so

that the reconstructive dictionaries are smoothly transformed into increasingly

discriminative representations. We also incorporate a new weighted sparse

coding scheme into our sparse recovery procedure, offering the system more

flexibility to adjust sparsity levels. Furthermore, we have devised a sparse

coding layer utilizing a ‘skinny’ dictionary. Integral to computational

efficiency, these skinny dictionaries compress the high dimensional sparse

codes into lower dimensional structures. The adaptivity and discriminability of

our 13-layer sparse coding network are demonstrated on four benchmark datasets,

namely Cifar-10, Cifar-100, SVHN and MNIST, most of which are considered

difficult for sparse coding models. Experimental results show that our

architecture overwhelmingly outperforms traditional one-layer sparse coding

architectures while using much fewer parameters. Moreover, our multilayer

architecture fuses the benefits of depth with sparse coding’s characteristic

ability to operate on smaller datasets. In such data-constrained scenarios, we

demonstrate our technique can overcome the limitations of deep neural networks

by exceeding the state of the art in accuracy.

Pooling Facial Segments to Face: The Shallow and Deep Ends

Generic face detection algorithms do not perform very well in the mobile

domain due to significant presence of occluded and partially visible faces. One

promising technique to handle the challenge of partial faces is to design face

detectors based on facial segments. In this paper two such face detectors

namely, SegFace and DeepSegFace, are proposed that detect the presence of a

face given arbitrary combinations of certain face segments. Both methods use

proposals from facial segments as input that are found using weak boosted

classifiers. SegFace is a shallow and fast algorithm using traditional

features, tailored for situations where real time constraints must be

satisfied. On the other hand, DeepSegFace is a more powerful algorithm based on

a deep convolutional neutral network (DCNN) architecture. DeepSegFace offers

certain advantages over other DCNN-based face detectors as it requires

relatively little amount of data to train by utilizing a novel data

augmentation scheme and is very robust to occlusion by design. Extensive

experiments show the superiority of the proposed methods, specially

DeepSegFace, over other state-of-the-art face detectors in terms of

precision-recall and ROC curve on two mobile face datasets.

Treelogy: A Novel Tree Classifier Utilizing Deep and Hand-crafted Representations

İlke Çuğu , Eren Şener , Çağrı Erciyes , Burak Balcı , Emre Akın , Itır Önal , Ahmet Oğuz Akyüz Subjects : Computer Vision and Pattern Recognition (cs.CV)

We propose a novel tree classification system called Treelogy, that fuses

deep representations with hand-crafted features obtained from leaf images to

perform leaf-based plant classification. Key to this system are segmentation of

the leaf from an untextured background, using convolutional neural networks

(CNNs) for learning deep representations, extracting hand-crafted features with

a number of image processing techniques, training a linear SVM with feature

vectors, merging SVM and CNN results, and identifying the species from a

dataset of 57 trees. Our classification results show that fusion of deep

representations with hand-crafted features leads to the highest accuracy. The

proposed algorithm is embedded in a smart-phone application, which is publicly

available. Furthermore, our novel dataset comprised of 5408 leaf images is also

made public for use of other researchers.

Face Detection using Deep Learning: An Improved Faster RCNN Approach

Xudong Sun , Pengcheng Wu , Steven C.H. Hoi Subjects : Computer Vision and Pattern Recognition (cs.CV)

In this report, we present a new face detection scheme using deep learning

and achieve the state-of-the-art detection performance on the well-known FDDB

face detetion benchmark evaluation. In particular, we improve the

state-of-the-art faster RCNN framework by combining a number of strategies,

including feature concatenation, hard negative mining, multi-scale training,

model pretraining, and proper calibration of key parameters. As a consequence,

the proposed scheme obtained the state-of-the-art face detection performance,

making it the best model in terms of ROC curves among all the published methods

on the FDDB benchmark.

Pruned non-local means

In Non-Local Means (NLM), each pixel is denoised by performing a weighted

averaging of its neighboring pixels, where the weights are computed using image

patches. We demonstrate that the denoising performance of NLM can be improved

by pruning the neighboring pixels, namely, by rejecting neighboring pixels

whose weights are below a certain threshold (lambda). While pruning can

potentially reduce pixel averaging in uniform-intensity regions, we demonstrate

that there is generally an overall improvement in the denoising performance. In

particular, the improvement comes from pixels situated close to edges and

corners. The success of the proposed method strongly depends on the choice of

the global threshold (lambda), which in turn depends on the noise level and

the image characteristics. We show how Stein’s unbiased estimator of the

mean-squared error can be used to optimally tune (lambda), at a marginal

computational overhead. We present some representative denoising results to

demonstrate the superior performance of the proposed method over NLM and its

variants.

Exploiting saliency for object segmentation from image level labels

There have been remarkable improvements in the semantic labelling task in the

recent years. However, the state of the art methods rely on large-scale

pixel-level annotations. This paper studies the problem of training a

pixel-wise semantic labeller network from image-level annotations of the

present object classes. Recently, it has been shown that high quality seeds

indicating discriminative object regions can be obtained from image-level

labels. Without additional information, obtaining the full extent of the object

is an inherently ill-posed problem due to co-occurrences. We propose using a

saliency model as additional information and hereby exploit prior knowledge on

the object extent and image statistics. We show how to combine both information

sources in order to recover 80% of the fully supervised performance – which is

the new state of the art in weakly supervised training for pixel-wise semantic

labelling.

Detection, Segmentation and Recognition of Face and its Features Using Neural Network

Face detection and recognition has been prevalent with research scholars and

diverse approaches have been incorporated till date to serve purpose. The

rampant advent of biometric analysis systems, which may be full body scanners,

or iris detection and recognition systems and the finger print recognition

systems, and surveillance systems deployed for safety and security purposes

have contributed to inclination towards same. Advances has been made with

frontal view, lateral view of the face or using facial expressions such as

anger, happiness and gloominess, still images and video image to be used for

detection and recognition. This led to newer methods for face detection and

recognition to be introduced in achieving accurate results and economically

feasible and extremely secure. Techniques such as Principal Component analysis

(PCA), Independent component analysis (ICA), Linear Discriminant Analysis

(LDA), have been the predominant ones to be used. But with improvements needed

in the previous approaches Neural Networks based recognition was like boon to

the industry. It not only enhanced the recognition but also the efficiency of

the process. Choosing Backpropagation as the learning method was clearly out of

its efficiency to recognize nonlinear faces with an acceptance ratio of more

than 90% and execution time of only few seconds.

Detection of Face using Viola Jones and Recognition using Back Propagation Neural Network

Detection and recognition of the facial images of people is an intricate

problem which has garnered much attention during recent years due to its ever

increasing applications in numerous fields. It continues to pose a challenge in

finding a robust solution to it. Its scope extends to catering the security,

commercial and law enforcement applications. Research for moreover a decade on

this subject has brought about remarkable development with the modus operandi

like human computer interaction, biometric analysis and content based coding of

images, videos and surveillance. A trivial task for brain but cumbersome to be

imitated artificially. The commonalities in faces does pose a problem on

various grounds but features such as skin color, gender differentiate a person

from the other. In this paper the facial detection has been carried out using

Viola-Jones algorithm and recognition of face has been done using Back

Propagation Neural Network (BPNN).

An Efficient Algebraic Solution to the Perspective-Three-Point Problem

Tong Ke , Stergios Roumeliotis Subjects : Computer Vision and Pattern Recognition (cs.CV)

In this work, we present an algebraic solution to the classical

perspective-3-point (P3P) problem for determining the position and attitude of

a camera from observations of three known reference points. In contrast to

previous approaches, we first directly determine the camera’s attitude by

employing the corresponding geometric constraints to formulate a system of

trigonometric equations. This is then efficiently solved, following an

algebraic approach, to determine the unknown rotation matrix and subsequently

the camera’s position. As compared to recent alternatives, our method avoids

computing unnecessary (and potentially numerically unstable) intermediate

results, and thus achieves higher numerical accuracy and robustness at a lower

computational cost. These benefits are validated through extensive Monte-Carlo

simulations for both nominal and close-to-singular geometric configurations.

Camera-Trap Images Segmentation using Multi-Layer Robust Principal Component Analysis

Camera trapping is a technique to study wildlife using automatic triggered

cameras. However, camera trapping collects a lot of false positives (images

without animals), which must be segmented before the classification step. This

paper presents a Multi-Layer Robust Principal Component Analysis (RPCA) for

camera-trap images segmentation. Our Multi-Layer RPCA uses histogram

equalization and Gaussian filter as pre-processing, texture and color

descriptors as features, and morphological filters with active contour as

post-processing. The experiments focus on computing the sparse and low-rank

matrices with different amounts of camera-trap images. We tested the

Multi-Layer RPCA in our camera-trap database. To our best knowledge, this paper

is the first work proposing Multi-Layer RPCA and using it for camera-trap

images segmentation.

Peduncle Detection of Sweet Pepper for Autonomous Crop Harvesting – Combined Colour and 3D Information

This paper presents a 3D visual detection method for the challenging task of

detecting peduncles of sweet peppers (Capsicum annuum) in the field. Cutting

the peduncle cleanly is one of the most difficult stages of the harvesting

process, where the peduncle is the part of the crop that attaches it to the

main stem of the plant. Accurate peduncle detection in 3D space is therefore a

vital step in reliable autonomous harvesting of sweet peppers, as this can lead

to precise cutting while avoiding damage to the surrounding plant. This paper

makes use of both colour and geometry information acquired from an RGB-D sensor

and utilises a supervised-learning approach for the peduncle detection task.

The performance of the proposed method is demonstrated and evaluated using

qualitative and quantitative results (the Area-Under-the-Curve (AUC) of the

detection precision-recall curve). We are able to achieve an AUC of 0.71 for

peduncle detection on field-grown sweet peppers. We release a set of manually

annotated 3D sweet pepper and peduncle images to assist the research community

in performing further research on this topic.

SafeDrive: A Robust Lane Tracking System for Autonomous and Assisted Driving Under Limited Visibility

Junaed Sattar , Jiawei Mo Subjects : Robotics (cs.RO) ; Computer Vision and Pattern Recognition (cs.CV)

We present an approach towards robust lane tracking for assisted and

autonomous driving, particularly under poor visibility. Autonomous detection of

lane markers improves road safety, and purely visual tracking is desirable for

widespread vehicle compatibility and reducing sensor intrusion, cost, and

energy consumption. However, visual approaches are often ineffective because of

a number of factors, including but not limited to occlusion, poor weather

conditions, and paint wear-off. Our method, named SafeDrive, attempts to

improve visual lane detection approaches in drastically degraded visual

conditions without relying on additional active sensors. In scenarios where

visual lane detection algorithms are unable to detect lane markers, the

proposed approach uses location information of the vehicle to locate and access

alternate imagery of the road and attempts detection on this secondary image.

Subsequently, by using a combination of feature-based and pixel-based

alignment, an estimated location of the lane marker is found in the current

scene. We demonstrate the effectiveness of our system on actual driving data

from locations in the United States with Google Street View as the source of

alternate imagery.

Transformation-Based Models of Video Sequences

Joost van Amersfoort , Anitha Kannan , Marc'Aurelio Ranzato , Arthur Szlam , Du Tran , Soumith Chintala Subjects : Learning (cs.LG) ; Computer Vision and Pattern Recognition (cs.CV)

In this work we propose a simple unsupervised approach for next frame

prediction in video. Instead of directly predicting the pixels in a frame given

past frames, we predict the transformations needed for generating the next

frame in a sequence, given the transformations of the past frames. This leads

to sharper results, while using a smaller prediction model.

In order to enable a fair comparison between different video frame prediction

models, we also propose a new evaluation protocol. We use generated frames as

input to a classifier trained with ground truth sequences. This criterion

guarantees that models scoring high are those producing sequences which

preserve discrim- inative features, as opposed to merely penalizing any

deviation, plausible or not, from the ground truth. Our proposed approach

compares favourably against more sophisticated ones on the UCF-101 data set,

while also being more efficient in terms of the number of parameters and

computational cost.

When Slepian Meets Fiedler: Putting a Focus on the Graph Spectrum

Network models play an important role in studying complex systems in many

scientific disciplines. Graph signal processing is receiving growing interest

as to design novel tools to combine the analysis of topology and signals. The

graph Fourier transform, defined as the eigendecomposition of the graph

Laplacian, allows extending conventional signal-processing operations to

graphs. One main feature is to let emerge global organization from local

interactions; i.e., the Fiedler vector has the smallest non-zero eigenvalue and

is key for Laplacian embedding and graph clustering. Here, we introduce the

design of Slepian graph signals, by maximizing energy concentration in a

predefined subgraph for a given spectral bandlimit. We also establish a link

with classical Laplacian embedding and graph clustering, for which the graph

Slepian design can serve as a generalization.

Image-Grounded Conversations: Multimodal Context for Natural Question and Response Generation

Nasrin Mostafazadeh , Chris Brockett , Bill Dolan , Michel Galley , Jianfeng Gao , Georgios P. Spithourakis , Lucy Vanderwende Subjects : Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)

The popularity of image sharing on social media reflects the important role

visual context plays in everyday conversation. In this paper, we present a

novel task, Image-Grounded Conversations (IGC), in which natural-sounding

conversations are generated about shared photographic images. We investigate

this task using training data derived from image-grounded conversations on

social media and introduce a new dataset of crowd-sourced conversations for

benchmarking progress. Experiments using deep neural network models trained on

social media data show that the combination of visual and textual context can

enhance the quality of generated conversational turns. In human evaluation, a

gap between human performance and that of both neural and retrieval

architectures suggests that IGC presents an interesting challenge for vision

and language research.

Sampling Without Time: Recovering Echoes of Light via Temporal Phase Retrieval

This paper considers the problem of sampling and reconstruction of a

continuous-time sparse signal without assuming the knowledge of the sampling

instants or the sampling rate. This topic has its roots in the problem of

recovering multiple echoes of light from its low-pass filtered and

auto-correlated, time-domain measurements. Our work is closely related to the

topic of sparse phase retrieval and in this context, we discuss the advantage

of phase-free measurements. While this problem is ill-posed, cues based on

physical constraints allow for its appropriate regularization. We validate our

theory with experiments based on customized, optical time-of-flight imaging

sensors. What singles out our approach is that our sensing method allows for

temporal phase retrieval as opposed to the usual case of spatial phase

retrieval. Preliminary experiments and results demonstrate a compelling

capability of our phase-retrieval based imaging device.

Artificial Intelligence

Diversification Methods for Zero-One Optimization

We introduce new diversification methods for zero-one optimization that

significantly extend strategies previously introduced in the setting of

metaheuristic search. Our methods incorporate easily implemented strategies for

partitioning assignments of values to variables, accompanied by processes

called augmentation and shifting which create greater flexibility and

generality. We then show how the resulting collection of diversified solutions

can be further diversified by means of permutation mappings, which equally can

be used to generate diversified collections of permutations for applications

such as scheduling and routing. These methods can be applied to non-binary

vectors by the use of binarization procedures and by Diversification-Based

Learning (DBL) procedures which also provide connections to applications in

clustering and machine learning. Detailed pseudocode and numerical

illustrations are provided to show the operation of our methods and the

collections of solutions they create.

Redefinition of the concept of fuzzy set based on vague partition from the perspective of axiomatization

Based on the in-depth analysis of the essence and features of vague

phenomena, this paper focuses on establishing the axiomatical foundation of

membership degree theory for vague phenomena, presents an axiomatic system to

govern membership degrees and their interconnections. On this basis, the

concept of vague partition is introduced, further, the concept of fuzzy set

introduced by Zadeh in 1965 is redefined based on vague partition from the

perspective of axiomatization. The thesis defended in this paper is that the

relationship among vague attribute values should be the starting point to

recognize and model vague phenomena from a quantitative view.

Credal Networks under Epistemic Irrelevance

Jasper De Bock Subjects : Artificial Intelligence (cs.AI) ; Probability (math.PR)

A credal network under epistemic irrelevance is a generalised type of

Bayesian network that relaxes its two main building blocks. On the one hand,

the local probabilities are allowed to be partially specified. On the other

hand, the assessments of independence do not have to hold exactly.

Conceptually, these two features turn credal networks under epistemic

irrelevance into a powerful alternative to Bayesian networks, offering a more

flexible approach to graph-based multivariate uncertainty modelling. However,

in practice, they have long been perceived as very hard to work with, both

theoretically and computationally.

The aim of this paper is to demonstrate that this perception is no longer

justified. We provide a general introduction to credal networks under epistemic

irrelevance, give an overview of the state of the art, and present several new

theoretical results. Most importantly, we explain how these results can be

combined to allow for the design of recursive inference methods. We provide

numerous concrete examples of how this can be achieved, and use these to

demonstrate that computing with credal networks under epistemic irrelevance is

most definitely feasible, and in some cases even highly efficient. We also

discuss several philosophical aspects, including the lack of symmetry, how to

deal with probability zero, the interpretation of lower expectations, the

axiomatic status of graphoid properties, and the difference between updating

and conditioning.

Survey on Models and Techniques for Root-Cause Analysis

Automation and computer intelligence to support complex human decisions

becomes essential to manage large and distributed systems in the Cloud and IoT

era. Understanding the root cause of an observed symptom in a complex system

has been a major problem for decades. As industry dives into the IoT world and

the amount of data generated per year grows at an amazing speed, an important

question is how to find appropriate mechanisms to determine root causes that

can handle huge amounts of data or may provide valuable feedback in real-time.

While many survey papers aim at summarizing the landscape of techniques for

modelling system behavior and infering the root cause of a problem based in the

resulting models, none of those focuses on analyzing how the different

techniques in the literature fit growing requirements in terms of performance

and scalability. In this survey, we provide a review of root-cause analysis,

focusing on these particular aspects. We also provide guidance to choose the

best root-cause analysis strategy depending on the requirements of a particular

system and application.

Rhythm Transcription of Polyphonic Piano Music Based on Merged-Output HMM for Multiple Voices

In a recent conference paper, we have reported a rhythm transcription method

based on a merged-output hidden Markov model (HMM) that explicitly describes

the multiple-voice structure of polyphonic music. This model solves a major

problem of conventional methods that could not properly describe the nature of

multiple voices as in polyrhythmic scores or in the phenomenon of loose

synchrony between voices. In this paper we present a complete description of

the proposed model and develop an inference technique, which is valid for any

merged-output HMMs for which output probabilities depend on past events. We

also examine the influence of the architecture and parameters of the method in

terms of accuracies of rhythm transcription and voice separation and perform

comparative evaluations with six other algorithms. Using MIDI recordings of

classical piano pieces, we found that the proposed model outperformed other

methods by more than 12 points in the accuracy for polyrhythmic performances

and performed almost as good as the best one for non-polyrhythmic performances.

This reveals the state-of-the-art methods of rhythm transcription for the first

time in the literature. Publicly available source codes are also provided for

future comparisons.

Explanation Generation as Model Reconciliation in Multi-Model Planning

Tathagata Chakraborti , Sarath Sreedharan , Yu Zhang , Subbarao Kambhampati Subjects : Artificial Intelligence (cs.AI)

The ability to explain the rationale behind a planner’s deliberative process

is crucial to the realization of effective human-planner interaction. However,

in the context of human-in-the-loop planning, a significant challenge towards

providing meaningful explanations arises due to the fact that the actor

(planner) and the observer (human) are likely to have different models of the

world, leading to a difference in the expected plan for the same perceived

planning problem. In this paper, for the first time, we formalize this notion

of Multi-Model Planning (MMP) and describe how a planner can provide

explanations of its plans in the context of such model differences.

Specifically, we will pose the multi-model explanation generation problem as a

model reconciliation problem and show how meaningful explanations may be

affected by making corrections to the human model. We will also demonstrate the

efficacy of our approach in randomly generated problems from benchmark planning

domains, and motivate exciting avenues of future research in the MMP paradigm.

Practical Reasoning with Norms for Autonomous Software Agents (Full Edition)

Zohreh Shams , Marina De Vos , Julian Padget , Wamberto W. Vasconcelos Subjects : Artificial Intelligence (cs.AI)

Autonomous software agents operating in dynamic environments need to

constantly reason about actions in pursuit of their goals, while taking into

consideration norms which might be imposed on those actions. Normative

practical reasoning supports agents making decisions about what is best for

them to (not) do in a given situation. What makes practical reasoning

challenging is the interplay between goals that agents are pursuing and the

norms that the agents are trying to uphold. We offer a formalisation to allow

agents to plan for multiple goals and norms in the presence of durative actions

that can be executed concurrently. We compare plans based on decision-theoretic

notions (i.e. utility) such that the utility gain of goals and utility loss of

norm violations are the basis for this comparison. The set of optimal plans

consists of plans that maximise the overall utility, each of which can be

chosen by the agent to execute. We provide an implementation of our proposal in

Answer Set Programming, thus allowing us to state the original problem in terms

of a logic program that can be queried for solutions with specific properties.

The implementation is proven to be sound and complete.

Multiclass MinMax Rank Aggregation

Pan Li , Olgica Milenkovic Subjects : Artificial Intelligence (cs.AI)

We introduce a new family of minmax rank aggregation problems under two

distance measures, the Kendall { au} and the Spearman footrule. As the

problems are NP-hard, we proceed to describe a number of constant-approximation

algorithms for solving them. We conclude with illustrative applications of the

aggregation methods on the Mallows model and genomic data.

A Study of FOSS'2013 Survey Data Using Clustering Techniques

FOSS is an acronym for Free and Open Source Software. The FOSS 2013 survey

primarily targets FOSS contributors and relevant anonymized dataset is publicly

available under CC by SA license. In this study, the dataset is analyzed from a

critical perspective using statistical and clustering techniques (especially

multiple correspondence analysis) with a strong focus on women contributors

towards discovering hidden trends and facts. Important inferences are drawn

about development practices and other facets of the free software and OSS

worlds.

Pure Rough Mereology and Counting

The study of mereology (parts and wholes) in the context of formal approaches

to vagueness can be approached in a number of ways. In the context of rough

sets, mereological concepts with a set-theoretic or valuation based ontology

acquire complex and diverse behavior. In this research a general rough set

framework called granular operator spaces is extended and the nature of

parthood in it is explored from a minimally intrusive point of view. This is

used to develop counting strategies that help in classifying the framework. The

developed methodologies would be useful for drawing involved conclusions about

the nature of data (and validity of assumptions about it) from antichains

derived from context. The problem addressed is also about whether counting

procedures help in confirming that the approximations involved in formation of

data are indeed rough approximations?

Incremental Maintenance Of Association Rules Under Support Threshold Change

Mohamed Anis Bach Tobji , Mohamed Salah Gouider Subjects : Artificial Intelligence (cs.AI) ; Databases (cs.DB)

Maintenance of association rules is an interesting problem. Several

incremental maintenance algorithms were proposed since the work of (Cheung et

al, 1996). The majority of these algorithms maintain rule bases assuming that

support threshold doesn’t change. In this paper, we present incremental

maintenance algorithm under support threshold change. This solution allows user

to maintain its rule base under any support threshold.

Comparative Study Of Data Mining Query Languages

Mohamed Anis Bach Tobji Subjects : Artificial Intelligence (cs.AI) ; Databases (cs.DB)

Since formulation of Inductive Database (IDB) problem, several Data Mining

(DM) languages have been proposed, confirming that KDD process could be

supported via inductive queries (IQ) answering. This paper reviews the existing

DM languages. We are presenting important primitives of the DM language and

classifying our languages according to primitives’ satisfaction. In addition,

we presented languages’ syntaxes and tried to apply each one to a database

sample to test a set of KDD operations. This study allows us to highlight

languages capabilities and limits, which is very useful for future work and

perspectives.

Methodologies for realizing natural-language-facilitated human-robot cooperation: A review

Natural Language (NL) for transferring knowledge from a human to a robot.

Recently, research on using NL to support human-robot cooperation (HRC) has

received increasing attention in several domains such as robotic daily

assistance, robotic health caregiving, intelligent manufacturing, autonomous

navigation and robot social accompany. However, a high-level review that can

reveal the realization process and the latest methodologies of using NL to

facilitate HRC is missing. In this review, a comprehensive summary about the

methodology development of natural-language-facilitated human-robot cooperation

(NLC) has been made. We first analyzed driving forces for NLC developments.

Then, with a temporal realization order, we reviewed three main steps of NLC:

human NL understanding, knowledge representation, and knowledge-world mapping.

Last, based on our paper review and perspectives, potential research trends in

NLC was discussed.

Decision structure of risky choice

As we know, there is a controversy about the decision making under risk

between economists and psychologists. We discuss to build a unified theory of

risky choice, which would explain both of compensatory and non-compensatory

theories. Obviously, decision strategy is not stuck in a rut, but based on the

things, in the real life, and experiment materials, in the laboratory. We

believe that human has a decision structure, which has constant and variable,

interval, concepts of probability and value. Namely, according to cognition

ability, we argue that people could not build a continuous and accurate

subjective probability world, but several intervals of probability perception.

More precisely, decision making is an order reduction process, which is

simplifying the decision structure. However, we are not really sure which

reduction path will occur during decision making process. It is why preference

reversal always happens when making decisions. The most efficient way to reduce

the order of decision structure is mathematical expectation. We also argue that

the deliberation time at least has four parts, which are consist of

substitution time,{ au}”(G) d{ au} time, { au}'(G) d{ au} time and

calculation time. Decision structure can simply explain the phenomenon of

paradoxes and anomalies. JEL Codes: C10, D03, D81.

Feature base fusion for splicing forgery detection based on neuro fuzzy

Habib Ghaffari Hadigheh , Ghazali bin sulong Subjects : Computer Vision and Pattern Recognition (cs.CV) ; Artificial Intelligence (cs.AI); Learning (cs.LG)

Most of researches on image forensics have been mainly focused on detection

of artifacts introduced by a single processing tool. They lead in the

development of many specialized algorithms looking for one or more particular

footprints under specific settings. Naturally, the performance of such

algorithms are not perfect, and accordingly the provided output might be noisy,

inaccurate and only partially correct. Furthermore, a forged image in practical

scenarios is often the result of utilizing several tools available by

image-processing software systems. Therefore, reliable tamper detection

requires developing more poweful tools to deal with various tempering

scenarios. Fusion of forgery detection tools based on Fuzzy Inference System

has been used before for addressing this problem. Adjusting the membership

functions and defining proper fuzzy rules for attaining to better results are

time-consuming processes. This can be accounted as main disadvantage of fuzzy

inference systems. In this paper, a Neuro-Fuzzy inference system for fusion of

forgery detection tools is developed. The neural network characteristic of

these systems provides appropriate tool for automatically adjusting the

membership functions. Moreover, initial fuzzy inference system is generated

based on fuzzy clustering techniques. The proposed framework is implemented and

validated on a benchmark image splicing data set in which three forgery

detection tools are fused based on adaptive Neuro-Fuzzy inference system. The

outcome of the proposed method reveals that applying Neuro Fuzzy inference

systems could be a better approach for fusion of forgery detection tools.

Systems of natural-language-facilitated human-robot cooperation: A review

Natural-language-facilitated human-robot cooperation (NLC), in which natural

language (NL) is used to share knowledge between a human and a robot for

conducting intuitive human-robot cooperation (HRC), is continuously developing

in the recent decade. Currently, NLC is used in several robotic domains such as

manufacturing, daily assistance and health caregiving. It is necessary to

summarize current NLC-based robotic systems and discuss the future developing

trends, providing helpful information for future NLC research. In this review,

we first analyzed the driving forces behind the NLC research. Regarding to a

robot s cognition level during the cooperation, the NLC implementations then

were categorized into four types {NL-based control, NL-based robot training,

NL-based task execution, NL-based social companion} for comparison and

discussion. Last based on our perspective and comprehensive paper review, the

future research trends were discussed.

Entropic Causality and Greedy Minimum Entropy Coupling

We study the problem of identifying the causal relationship between two

discrete random variables from observational data. We recently proposed a novel

framework called entropic causality that works in a very general functional

model but makes the assumption that the unobserved exogenous variable has small

entropy in the true causal direction.

This framework requires the solution of a minimum entropy coupling problem:

Given marginal distributions of m discrete random variables, each on n states,

find the joint distribution with minimum entropy, that respects the given

marginals. This corresponds to minimizing a concave function of nm variables

over a convex polytope defined by nm linear constraints, called a

transportation polytope. Unfortunately, it was recently shown that this minimum

entropy coupling problem is NP-hard, even for 2 variables with n states. Even

representing points (joint distributions) over this space can require

exponential complexity (in n, m) if done naively.

In our recent work we introduced an efficient greedy algorithm to find an

approximate solution for this problem. In this paper we analyze this algorithm

and establish two results: that our algorithm always finds a local minimum and

also is within an additive approximation error from the unknown global optimum.

Image-Grounded Conversations: Multimodal Context for Natural Question and Response Generation

Nasrin Mostafazadeh , Chris Brockett , Bill Dolan , Michel Galley , Jianfeng Gao , Georgios P. Spithourakis , Lucy Vanderwende Subjects : Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)

The popularity of image sharing on social media reflects the important role

visual context plays in everyday conversation. In this paper, we present a

novel task, Image-Grounded Conversations (IGC), in which natural-sounding

conversations are generated about shared photographic images. We investigate

this task using training data derived from image-grounded conversations on

social media and introduce a new dataset of crowd-sourced conversations for

benchmarking progress. Experiments using deep neural network models trained on

social media data show that the combination of visual and textual context can

enhance the quality of generated conversational turns. In human evaluation, a

gap between human performance and that of both neural and retrieval

architectures suggests that IGC presents an interesting challenge for vision

and language research.

Information Retrieval

Click Through Rate Prediction for Contextual Advertisment Using Linear Regression

This research presents an innovative and unique way of solving the

advertisement prediction problem which is considered as a learning problem over

the past several years. Online advertising is a multi-billion-dollar industry

and is growing every year with a rapid pace. The goal of this research is to

enhance click through rate of the contextual advertisements using Linear

Regression. In order to address this problem, a new technique propose in this

paper to predict the CTR which will increase the overall revenue of the system

by serving the advertisements more suitable to the viewers with the help of

feature extraction and displaying the advertisements based on context of the

publishers. The important steps include the data collection, feature

extraction, CTR prediction and advertisement serving. The statistical results

obtained from the dynamically used technique show an efficient outcome by

fitting the data close to perfection for the LR technique using optimized

feature selection.

Feature Studies to Inform the Classification of Depressive Symptoms from Twitter Data for Population Health

Danielle Mowery , Craig Bryan , Mike Conway Subjects : Information Retrieval (cs.IR) ; Computation and Language (cs.CL); Computers and Society (cs.CY); Social and Information Networks (cs.SI)

The utility of Twitter data as a medium to support population-level mental

health monitoring is not well understood. In an effort to better understand the

predictive power of supervised machine learning classifiers and the influence

of feature sets for efficiently classifying depression-related tweets on a

large-scale, we conducted two feature study experiments. In the first

experiment, we assessed the contribution of feature groups such as lexical

information (e.g., unigrams) and emotions (e.g., strongly negative) using a

feature ablation study. In the second experiment, we determined the percentile

of top ranked features that produced the optimal classification performance by

applying a three-step feature elimination approach. In the first experiment, we

observed that lexical features are critical for identifying depressive

symptoms, specifically for depressed mood (-35 points) and for disturbed sleep

(-43 points). In the second experiment, we observed that the optimal F1-score

performance of top ranked features in percentiles variably ranged across

classes e.g., fatigue or loss of energy (5th percentile, 288 features) to

depressed mood (55th percentile, 3,168 features) suggesting there is no

consistent count of features for predicting depressive-related tweets. We

conclude that simple lexical features and reduced feature sets can produce

comparable results to larger feature sets.

Binary adaptive embeddings from order statistics of random projections

Diego Valsesia , Enrico Magli Subjects : Learning (cs.LG) ; Information Retrieval (cs.IR)

We use some of the largest order statistics of the random projections of a

reference signal to construct a binary embedding that is adapted to signals

correlated with such signal. The embedding is characterized from the analytical

standpoint and shown to provide improved performance on tasks such as

classification in a reduced-dimensionality space.

Who With Whom And How?: Extracting Large Social Networks Using Search Engines

Social network analysis is leveraged in a variety of applications such as

identifying influential entities, detecting communities with special interests,

and determining the flow of information and innovations. However, existing

approaches for extracting social networks from unstructured Web content do not

scale well and are only feasible for small graphs. In this paper, we introduce

novel methodologies for query-based search engine mining, enabling efficient

extraction of social networks from large amounts of Web data. To this end, we

use patterns in phrase queries for retrieving entity connections, and employ a

bootstrapping approach for iteratively expanding the pattern set. Our

experimental evaluation in different domains demonstrates that our algorithms

provide high quality results and allow for scalable and efficient construction

of social graphs.

How to Search the Internet Archive Without Indexing It

Significant parts of cultural heritage are produced on the web during the

last decades. While easy accessibility to the current web is a good baseline,

optimal access to the past web faces several challenges. This includes dealing

with large-scale web archive collections and lacking of usage logs that contain

implicit human feedback most relevant for today’s web search. In this paper, we

propose an entity-oriented search system to support retrieval and analytics on

the Internet Archive. We use Bing to retrieve a ranked list of results from the

current web. In addition, we link retrieved results to the WayBack Machine;

thus allowing keyword search on the Internet Archive without processing and

indexing its raw archived content. Our search system complements existing web

archive search tools through a user-friendly interface, which comes close to

the functionalities of modern web search engines (e.g., keyword search, query

auto-completion and related query suggestion), and provides a great benefit of

taking user feedback on the current web into account also for web archive

search. Through extensive experiments, we conduct quantitative and qualitative

analyses in order to provide insights that enable further research on and

practical applications of web archives.

Computation and Language

Bangla Word Clustering Based on Tri-gram, 4-gram and 5-gram Language Model

In this paper, we describe a research method that generates Bangla word

clusters on the basis of relating to meaning in language and contextual

similarity. The importance of word clustering is in parts of speech (POS)

tagging, word sense disambiguation, text classification, recommender system,

spell checker, grammar checker, knowledge discover and for many others Natural

Language Processing (NLP) applications. In the history of word clustering,

English and some other languages have already implemented some methods on word

clustering efficiently. But due to lack of the resources, word clustering in

Bangla has not been still implemented efficiently. Presently, its

implementation is in the beginning stage. In some research of word clustering

in English based on preceding and next five words of a key word they found an

efficient result. Now, we are trying to implement the tri-gram, 4-gram and

5-gram model of word clustering for Bangla to observe which one is the best

among them. We have started our research with quite a large corpus of

approximate 1 lakh Bangla words. We are using a machine learning technique in

this research. We will generate word clusters and analyze the clusters by

testing some different threshold values.

A Comparative Study on Different Types of Approaches to Bengali document Categorization

Document categorization is a technique where the category of a document is

determined. In this paper three well-known supervised learning techniques which

are Support Vector Machine(SVM), Na”ive Bayes(NB) and Stochastic Gradient

Descent(SGD) compared for Bengali document categorization. Besides classifier,

classification also depends on how feature is selected from dataset. For

analyzing those classifier performances on predicting a document against twelve

categories several feature selection techniques are also applied in this

article namely Chi square distribution, normalized TFIDF (term

frequency-inverse document frequency) with word analyzer. So, we attempt to

explore the efficiency of those three-classification algorithms by using two

different feature selection techniques in this article.

Structural Analysis of Hindi Phonetics and A Method for Extraction of Phonetically Rich Sentences from a Very Large Hindi Text Corpus

Automatic speech recognition (ASR) and Text to speech (TTS) are two prominent

area of research in human computer interaction nowadays. A set of phonetically

rich sentences is in a matter of importance in order to develop these two

interactive modules of HCI. Essentially, the set of phonetically rich sentences

has to cover all possible phone units distributed uniformly. Selecting such a

set from a big corpus with maintaining phonetic characteristic based similarity

is still a challenging problem. The major objective of this paper is to devise

a criteria in order to select a set of sentences encompassing all phonetic

aspects of a corpus with size as minimum as possible. First, this paper

presents a statistical analysis of Hindi phonetics by observing the structural

characteristics. Further a two stage algorithm is proposed to extract

phonetically rich sentences with a high variety of triphones from the EMILLE

Hindi corpus. The algorithm consists of a distance measuring criteria to select

a sentence in order to improve the triphone distribution. Moreover, a special

preprocessing method is proposed to score each triphone in terms of inverse

probability in order to fasten the algorithm. The results show that the

approach efficiently build uniformly distributed phonetically-rich corpus with

optimum number of sentences.

Graph-Based Semi-Supervised Conditional Random Fields For Spoken Language Understanding Using Unaligned Data

We experiment graph-based Semi-Supervised Learning (SSL) of Conditional

Random Fields (CRF) for the application of Spoken Language Understanding (SLU)

on unaligned data. The aligned labels for examples are obtained using IBM

Model. We adapt a baseline semi-supervised CRF by defining new feature set and

altering the label propagation algorithm. Our results demonstrate that our

proposed approach significantly improves the performance of the supervised

model by utilizing the knowledge gained from the graph.

Extracting Bilingual Persian Italian Lexicon from Comparable Corpora Using Different Types of Seed Dictionaries

Bilingual dictionaries are very important in various fields of natural

language processing. In recent years, research on extracting new bilingual

lexicons from non-parallel (comparable) corpora have been proposed. Almost all

use a small existing dictionary or other resource to make an initial list

called the “seed dictionary”. In this paper we discuss the use of different

types of dictionaries as the initial starting list for creating a bilingual

Persian-Italian lexicon from a comparable corpus.

Our experiments apply state-of-the-art techniques on three different seed

dictionaries; an existing dictionary, a dictionary created with pivot-based

schema, and a dictionary extracted from a small Persian-Italian parallel text.

The interesting challenge of our approach is to find a way to combine different

dictionaries together in order to produce a better and more accurate lexicon.

In order to combine seed dictionaries, we propose two different combination

models and examine the effect of our novel combination models on various

comparable corpora that have differing degrees of comparability. We conclude

with a proposal for a new weighting system to improve the extracted lexicon.

The experimental results produced by our implementation show the efficiency of

our proposed models.

Using English as Pivot to Extract Persian-Italian Parallel Sentences from Non-Parallel Corpora

The effectiveness of a statistical machine translation system (SMT) is very

dependent upon the amount of parallel corpus used in the training phase. For

low-resource language pairs there are not enough parallel corpora to build an

accurate SMT. In this paper, a novel approach is presented to extract bilingual

Persian-Italian parallel sentences from a non-parallel (comparable) corpus. In

this study, English is used as the pivot language to compute the matching

scores between source and target sentences and candidate selection phase.

Additionally, a new monolingual sentence similarity metric, Normalized Google

Distance (NGD) is proposed to improve the matching process. Moreover, some

extensions of the baseline system are applied to improve the quality of

extracted sentences measured with BLEU. Experimental results show that using

the new pivot based extraction can increase the quality of bilingual corpus

significantly and consequently improves the performance of the Persian-Italian

SMT system.

Drug-Drug Interaction Extraction from Biomedical Text Using Long Short Term Memory Network

A drug can affect the activity of other drugs, when administered together, in

both synergistic or antagonistic ways. In one hand synergistic effects lead to

improved therapeutic outcomes, antagonistic consequences can be

life-threatening, leading to increased healthcare cost, or may even cause

death. Thus, identification of unknown drug-drug interaction (DDI) is an

important concern for efficient and effective healthcare. Although there exist

multiple resources for DDI, they often unable to keep pace with rich amount of

information available in fast growing biomedical texts including literature.

Most existing methods model DDI extraction from text as classification problem

and mainly rely on handcrafted features. Some of these features further depends

on domain specific tools. Recently neural network models using latent features

has shown to be perform similar or better than the other existing models using

handcrafted features. In this paper, we present three models namely, B-LSTM,

AB-LSTM and Joint AB-LSTM based on long short-term memory (LSTM) network. All

three models utilize word and position embedding as latent features and thus do

not rely on feature engineering. Further use of bidirectional long short-term

memory (Bi-LSTM) networks allow to extract optimal features from the whole

sentence. The two models, AB-LSTM and Joint AB-LSTM also use attentive pooling

in the output of Bi-LSTM layer to assign weights to features. Our experimental

results on the SemEval-2013 DDI extraction dataset shows that the Joint AB-LSTM

model outperforms all the existing methods, including those relying on

handcrafted features. The other two proposed models also perform competitively

with state-of-the-art methods.

Image-Grounded Conversations: Multimodal Context for Natural Question and Response Generation

Nasrin Mostafazadeh , Chris Brockett , Bill Dolan , Michel Galley , Jianfeng Gao , Georgios P. Spithourakis , Lucy Vanderwende Subjects : Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)

The popularity of image sharing on social media reflects the important role

visual context plays in everyday conversation. In this paper, we present a

novel task, Image-Grounded Conversations (IGC), in which natural-sounding

conversations are generated about shared photographic images. We investigate

this task using training data derived from image-grounded conversations on

social media and introduce a new dataset of crowd-sourced conversations for

benchmarking progress. Experiments using deep neural network models trained on

social media data show that the combination of visual and textual context can

enhance the quality of generated conversational turns. In human evaluation, a

gap between human performance and that of both neural and retrieval

architectures suggests that IGC presents an interesting challenge for vision

and language research.

Adversarial Evaluation of Dialogue Models

Anjuli Kannan , Oriol Vinyals Subjects : Computation and Language (cs.CL)

The recent application of RNN encoder-decoder models has resulted in

substantial progress in fully data-driven dialogue systems, but evaluation

remains a challenge. An adversarial loss could be a way to directly evaluate

the extent to which generated dialogue responses sound like they came from a

human. This could reduce the need for human evaluation, while more directly

evaluating on a generative task. In this work, we investigate this idea by

training an RNN to discriminate a dialogue model’s samples from human-generated

samples. Although we find some evidence this setup could be viable, we also

note that many issues remain in its practical application. We discuss both

aspects and conclude that future work is warranted.

Methodologies for realizing natural-language-facilitated human-robot cooperation: A review

Natural Language (NL) for transferring knowledge from a human to a robot.

Recently, research on using NL to support human-robot cooperation (HRC) has

received increasing attention in several domains such as robotic daily

assistance, robotic health caregiving, intelligent manufacturing, autonomous

navigation and robot social accompany. However, a high-level review that can

reveal the realization process and the latest methodologies of using NL to

facilitate HRC is missing. In this review, a comprehensive summary about the

methodology development of natural-language-facilitated human-robot cooperation

(NLC) has been made. We first analyzed driving forces for NLC developments.

Then, with a temporal realization order, we reviewed three main steps of NLC:

human NL understanding, knowledge representation, and knowledge-world mapping.

Last, based on our paper review and perspectives, potential research trends in

NLC was discussed.

Document Decomposition of Bangla Printed Text

Today all kind of information is getting digitized and along with all this

digitization, the huge archive of various kinds of documents is being digitized

too. We know that, Optical Character Recognition is the method through which,

newspapers and other paper documents convert into digital resources. But, it is

a fact that this method works on texts only. As a result, if we try to process

any document which contains non-textual zones, then we will get garbage texts

as output. That is why; in order to digitize documents properly they should be

prepossessed carefully. And while preprocessing, segmenting document in

different regions according to the category properly is most important. But,

the Optical Character Recognition processes available for Bangla language have

no such algorithm that can categorize a newspaper/book page fully. So we worked

to decompose a document into its several parts like headlines, sub headlines,

columns, images etc. And if the input is skewed and rotated, then the input was

also deskewed and de-rotated. To decompose any Bangla document we found out the

edges of the input image. Then we find out the horizontal and vertical area of

every pixel where it lies in. Later on the input image was cut according to

these areas. Then we pick each and every sub image and found out their

height-width ratio, line height. Then according to these values the sub images

were categorized. To deskew the image we found out the skew angle and de skewed

the image according to this angle. To de-rotate the image we used the line

height, matra line, pixel ratio of matra line.

Systems of natural-language-facilitated human-robot cooperation: A review

Natural-language-facilitated human-robot cooperation (NLC), in which natural

language (NL) is used to share knowledge between a human and a robot for

conducting intuitive human-robot cooperation (HRC), is continuously developing

in the recent decade. Currently, NLC is used in several robotic domains such as

manufacturing, daily assistance and health caregiving. It is necessary to

summarize current NLC-based robotic systems and discuss the future developing

trends, providing helpful information for future NLC research. In this review,

we first analyzed the driving forces behind the NLC research. Regarding to a

robot s cognition level during the cooperation, the NLC implementations then

were categorized into four types {NL-based control, NL-based robot training,

NL-based task execution, NL-based social companion} for comparison and

discussion. Last based on our perspective and comprehensive paper review, the

future research trends were discussed.

Feature Studies to Inform the Classification of Depressive Symptoms from Twitter Data for Population Health

Danielle Mowery , Craig Bryan , Mike Conway Subjects : Information Retrieval (cs.IR) ; Computation and Language (cs.CL); Computers and Society (cs.CY); Social and Information Networks (cs.SI)

The utility of Twitter data as a medium to support population-level mental

health monitoring is not well understood. In an effort to better understand the

predictive power of supervised machine learning classifiers and the influence

of feature sets for efficiently classifying depression-related tweets on a

large-scale, we conducted two feature study experiments. In the first

experiment, we assessed the contribution of feature groups such as lexical

information (e.g., unigrams) and emotions (e.g., strongly negative) using a

feature ablation study. In the second experiment, we determined the percentile

of top ranked features that produced the optimal classification performance by

applying a three-step feature elimination approach. In the first experiment, we

observed that lexical features are critical for identifying depressive

symptoms, specifically for depressed mood (-35 points) and for disturbed sleep

(-43 points). In the second experiment, we observed that the optimal F1-score

performance of top ranked features in percentiles variably ranged across

classes e.g., fatigue or loss of energy (5th percentile, 288 features) to

depressed mood (55th percentile, 3,168 features) suggesting there is no

consistent count of features for predicting depressive-related tweets. We

conclude that simple lexical features and reduced feature sets can produce

comparable results to larger feature sets.

A Comprehensive Survey on Bengali Phoneme Recognition

Hidden Markov model based various phoneme recognition methods for Bengali

language is reviewed. Automatic phoneme recognition for Bengali language using

multilayer neural network is reviewed. Usefulness of multilayer neural network

over single layer neural network is discussed. Bangla phonetic feature table

construction and enhancement for Bengali speech recognition is also discussed.

Comparison among these methods is discussed.

Distributed, Parallel, and Cluster Computing

Fog-Assisted wIoT: A Smart Fog Gateway for End-to-End Analytics in Wearable Internet of Things

Today, wearable internet-of-things (wIoT) devices continuously flood the

cloud data centers at an enormous rate. This increases a demand to deploy an

edge infrastructure for computing, intelligence, and storage close to the

users. The emerging paradigm of fog computing could play an important role to

make wIoT more efficient and affordable. Fog computing is known as the cloud on

the ground. This paper presents an end-to-end architecture that performs data

conditioning and intelligent filtering for generating smart analytics from

wearable data. In wIoT, wearable sensor devices serve on one end while the

cloud backend offers services on the other end. We developed a prototype of

smart fog gateway (a middle layer) using Intel Edison and Raspberry Pi. We

discussed the role of the smart fog gateway in orchestrating the process of

data conditioning, intelligent filtering, smart analytics, and selective

transfer to the cloud for long-term storage and temporal variability

monitoring. We benchmarked the performance of developed prototypes on

real-world data from smart e-textile gloves. Results demonstrated the usability

and potential of proposed architecture for converting the real-world data into

useful analytics while making use of knowledge-based models. In this way, the

smart fog gateway enhances the end-to-end interaction between wearables (sensor

devices) and the cloud.

Autotuning GPU Kernels via Static and Predictive Analysis

Robert V. Lim , Boyana Norris , Allen D. Malony Subjects : Distributed, Parallel, and Cluster Computing (cs.DC) ; Performance (cs.PF)

Optimizing the performance of GPU kernels is challenging for both human

programmers and code generators. For example, CUDA programmers must set thread

and block parameters for a kernel, but might not have the intuition to make a

good choice. Similarly, compilers can generate working code, but may miss

tuning opportunities by not targeting GPU models or performing code

transformations. Although empirical autotuning addresses some of these

challenges, it requires extensive experimentation and search for optimal code

variants. This research presents an approach for tuning CUDA kernels based on

static analysis that considers fine-grained code structure and the specific GPU

architecture features. Notably, our approach does not require any program runs

in order to discover near-optimal parameter settings. We demonstrate the

applicability of our approach in enabling code autotuners such as Orio to

produce competitive code variants comparable with empirical-based methods,

without the high cost of experiments.

RIoTBench: A Real-time IoT Benchmark for Distributed Stream Processing Platforms

The Internet of Things (IoT) is an emerging technology paradigm where

millions of sensors and actuators help monitor and manage, physical,

environmental and human systems in real-time. The inherent closedloop

responsiveness and decision making of IoT applications make them ideal

candidates for using low latency and scalable stream processing platforms.

Distributed Stream Processing Systems (DSPS) hosted on Cloud data-centers are

becoming the vital engine for real-time data processing and analytics in any

IoT software architecture. But the efficacy and performance of contemporary

DSPS have not been rigorously studied for IoT applications and data streams.

Here, we develop RIoTBench, a Realtime IoT Benchmark suite, along with

performance metrics, to evaluate DSPS for streaming IoT applications. The

benchmark includes 27 common IoT tasks classified across various functional

categories and implemented as reusable micro-benchmarks. Further, we propose

four IoT application benchmarks composed from these tasks, and that leverage

various dataflow semantics of DSPS. The applications are based on common IoT

patterns for data pre-processing, statistical summarization and predictive

analytics. These are coupled with four stream workloads sourced from real IoT

observations on smart cities and fitness, with peak streams rates that range

from 500 to 10000 messages/sec and diverse frequency distributions. We validate

the RIoTBench suite for the popular Apache Storm DSPS on the Microsoft Azure

public Cloud, and present empirical observations. This suite can be used by

DSPS researchers for performance analysis and resource scheduling, and by IoT

practitioners to evaluate DSPS platforms.

IFCIoT: Integrated Fog Cloud IoT Architectural Paradigm for Future Internet of Things

We propose a novel integrated fog cloud IoT (IFCIoT) architectural paradigm

that promises increased performance, energy efficiency, reduced latency,

quicker response time, scalability, and better localized accuracy for future

IoT applications. The fog nodes (e.g., edge servers, smart routers, base

stations) receive computation offloading requests and sensed data from various

IoT devices. To enhance performance, energy efficiency, and real-time

responsiveness of applications, we propose a reconfigurable and layered fog

node (edge server) architecture that analyzes the applications’ characteristics

and reconfigure the architectural resources to better meet the peak workload

demands. The layers of the proposed fog node architecture include application

layer, analytics layer, virtualization layer, reconfiguration layer, and

hardware layer. The layered architecture facilitates abstraction and

implementation for fog computing paradigm that is distributed in nature and

where multiple vendors (e.g., applications, services, data and content

providers) are involved. We also elaborate the potential applications of IFCIoT

architecture, such as smart cities, intelligent transportation systems,

localized weather maps and environmental monitoring, and real-time agricultural

data analytics and control.

Accelerated Computing in Magnetic Resonance Imaging – Real-Time Imaging Using Non-Linear Inverse Reconstruction

Purpose: To develop generic optimization strategies for image reconstruction

using graphical processing units (GPUs) in magnetic resonance imaging (MRI) and

to exemplarily report about our experience with a highly accelerated

implementation of the non-linear inversion algorithm (NLINV) for dynamic MRI

with high frame rates. Methods: The NLINV algorithm is optimized and ported to

run on an a multi-GPU single-node server. The algorithm is mapped to multiple

GPUs by decomposing the data domain along the channel dimension. Furthermore,

the algorithm is decomposed along the temporal domain by relaxing a temporal

regularization constraint, allowing the algorithm to work on multiple frames in

parallel. Finally, an autotuning method is presented that is capable of

combining different decomposition variants to achieve optimal algorithm

performance in different imaging scenarios. Results: The algorithm is

successfully ported to a multi-GPU system and allows online image

reconstruction with high frame rates. Real-time reconstruction with low latency

and frame rates up to 30 frames per second is demonstrated. Conclusion: Novel

parallel decomposition methods are presented which are applicable to many

iterative algorithms for dynamic MRI. Using these methods to parallelize the

NLINV algorithm on multiple GPUs it is possible to achieve online image

reconstruction with high frame rates.

pMR: A high-performance communication library

On many parallel machines, the time LQCD applications spent in communication

is a significant contribution to the total wall-clock time, especially in the

strong-scaling limit. We present a novel high-performance communication library

that can be used as a de facto drop-in replacement for MPI in existing

software. Its lightweight nature that avoids some of the unnecessary overhead

introduced by MPI allows us to improve the communication performance of

applications without any algorithmic or complicated implementation changes. As

a first real-world benchmark, we make use of the pMR library in the coarse-grid

solve of the Regensburg implementation of the DD-(alpha)AMG algorithm. On

realistic lattices, we see an improvement of a factor 2x in pure communication

time and total execution time savings of up to 20%.

Learning

Memory Augmented Neural Networks with Wormhole Connections

Caglar Gulcehre , Sarath Chandar , Yoshua Bengio Subjects : Learning (cs.LG) ; Neural and Evolutionary Computing (cs.NE); Machine Learning (stat.ML)

Recent empirical results on long-term dependency tasks have shown that neural

networks augmented with an external memory can learn the long-term dependency

tasks more easily and achieve better generalization than vanilla recurrent

neural networks (RNN). We suggest that memory augmented neural networks can

reduce the effects of vanishing gradients by creating shortcut (or wormhole)

connections. Based on this observation, we propose a novel memory augmented

neural network model called TARDIS (Temporal Automatic Relation Discovery in

Sequences). The controller of TARDIS can store a selective set of embeddings of

its own previous hidden states into an external memory and revisit them as and

when needed. For TARDIS, memory acts as a storage for wormhole connections to

the past to propagate the gradients more effectively and it helps to learn the

temporal dependencies. The memory structure of TARDIS has similarities to both

Neural Turing Machines (NTM) and Dynamic Neural Turing Machines (D-NTM), but

both read and write operations of TARDIS are simpler and more efficient. We use

discrete addressing for read/write operations which helps to substantially to

reduce the vanishing gradient problem with very long sequences. Read and write

operations in TARDIS are tied with a heuristic once the memory becomes full,

and this makes the learning problem simpler when compared to NTM or D-NTM type

of architectures. We provide a detailed analysis on the gradient propagation in

general for MANNs. We evaluate our models on different long-term dependency

tasks and report competitive results in all of them.

Predicting Auction Price of Vehicle License Plate with Deep Recurrent Neural Network

Vinci Chow Subjects : Learning (cs.LG) ; Economics (q-fin.EC); Machine Learning (stat.ML)

In Chinese societies where superstition is of paramount importance, vehicle

license plates with desirable numbers can fetch for very high prices in

auctions. Unlike auctions of other valuable items, however, license plates do

not get an estimated price before auction. In this paper, I construct a deep

recurrent neural network to predict the prices of vehicle license plates in

Hong Kong based on the characters on a plate. Trained with 13-years of

historical auction prices, the deep RNN outperforms previous models by

significant margin.

A Unifying Framework for Guiding Point Processes with Stochastic Intensity Functions

Yichen Wang , Grady Williams , Evangelos Theodorou , Le Song Subjects : Learning (cs.LG) ; Social and Information Networks (cs.SI); Systems and Control (cs.SY); Optimization and Control (math.OC)

Temporal point processes are powerful tools to model event occurrences and

have a plethora of applications in social sciences. While the majority of prior

works focus on the modeling and learning of these processes, we consider the

problem of how to design the optimal control policy for general point process

with stochastic intensities, such that the stochastic system driven by the

process is steered to a target state. In particular, we exploit the novel

insight from the information theoretic formulations of stochastic optimal

control. We further propose a novel convex optimization framework and a highly

efficient online algorithm to update the policy adaptively to the current

system state. Experiments on synthetic and real-world data show that our

algorithm can steer the user activities much more accurately than

state-of-arts.

Binary adaptive embeddings from order statistics of random projections

Diego Valsesia , Enrico Magli Subjects : Learning (cs.LG) ; Information Retrieval (cs.IR)

We use some of the largest order statistics of the random projections of a

reference signal to construct a binary embedding that is adapted to signals

correlated with such signal. The embedding is characterized from the analytical

standpoint and shown to provide improved performance on tasks such as

classification in a reduced-dimensionality space.

Model-based Classification and Novelty Detection For Point Pattern Data

Point patterns are sets or multi-sets of unordered elements that can be found

in numerous data sources. However, in data analysis tasks such as

classification and novelty detection, appropriate statistical models for point

pattern data have not received much attention. This paper proposes the

modelling of point pattern data via random finite sets (RFS). In particular, we

propose appropriate likelihood functions, and a maximum likelihood estimator

for learning a tractable family of RFS models. In novelty detection, we propose

novel ranking functions based on RFS models, which substantially improve

performance.

Transformation-Based Models of Video Sequences

Joost van Amersfoort , Anitha Kannan , Marc'Aurelio Ranzato , Arthur Szlam , Du Tran , Soumith Chintala Subjects : Learning (cs.LG) ; Computer Vision and Pattern Recognition (cs.CV)

In this work we propose a simple unsupervised approach for next frame

prediction in video. Instead of directly predicting the pixels in a frame given

past frames, we predict the transformations needed for generating the next

frame in a sequence, given the transformations of the past frames. This leads

to sharper results, while using a smaller prediction model.

In order to enable a fair comparison between different video frame prediction

models, we also propose a new evaluation protocol. We use generated frames as

input to a classifier trained with ground truth sequences. This criterion

guarantees that models scoring high are those producing sequences which

preserve discrim- inative features, as opposed to merely penalizing any

deviation, plausible or not, from the ground truth. Our proposed approach

compares favourably against more sophisticated ones on the UCF-101 data set,

while also being more efficient in terms of the number of parameters and

computational cost.

When Slepian Meets Fiedler: Putting a Focus on the Graph Spectrum

Network models play an important role in studying complex systems in many

scientific disciplines. Graph signal processing is receiving growing interest

as to design novel tools to combine the analysis of topology and signals. The

graph Fourier transform, defined as the eigendecomposition of the graph

Laplacian, allows extending conventional signal-processing operations to

graphs. One main feature is to let emerge global organization from local

interactions; i.e., the Fiedler vector has the smallest non-zero eigenvalue and

is key for Laplacian embedding and graph clustering. Here, we introduce the

design of Slepian graph signals, by maximizing energy concentration in a

predefined subgraph for a given spectral bandlimit. We also establish a link

with classical Laplacian embedding and graph clustering, for which the graph

Slepian design can serve as a generalization.

Click Through Rate Prediction for Contextual Advertisment Using Linear Regression

This research presents an innovative and unique way of solving the

advertisement prediction problem which is considered as a learning problem over

the past several years. Online advertising is a multi-billion-dollar industry

and is growing every year with a rapid pace. The goal of this research is to

enhance click through rate of the contextual advertisements using Linear

Regression. In order to address this problem, a new technique propose in this

paper to predict the CTR which will increase the overall revenue of the system

by serving the advertisements more suitable to the viewers with the help of

feature extraction and displaying the advertisements based on context of the

publishers. The important steps include the data collection, feature

extraction, CTR prediction and advertisement serving. The statistical results

obtained from the dynamically used technique show an efficient outcome by

fitting the data close to perfection for the LR technique using optimized

feature selection.

PathNet: Evolution Channels Gradient Descent in Super Neural Networks

Chrisantha Fernando , Dylan Banarse , Charles Blundell , Yori Zwols , David Ha , Andrei A. Rusu , Alexander Pritzel , Daan Wierstra Subjects : Neural and Evolutionary Computing (cs.NE) ; Learning (cs.LG)

For artificial general intelligence (AGI) it would be efficient if multiple

users trained the same giant neural network, permitting parameter reuse,

without catastrophic forgetting. PathNet is a first step in this direction. It

is a neural network algorithm that uses agents embedded in the neural network

whose task is to discover which parts of the network to re-use for new tasks.

Agents are pathways (views) through the network which determine the subset of

parameters that are used and updated by the forwards and backwards passes of

the backpropogation algorithm. During learning, a tournament selection genetic

algorithm is used to select pathways through the neural network for replication

and mutation. Pathway fitness is the performance of that pathway measured

according to a cost function. We demonstrate successful transfer learning;

fixing the parameters along a path learned on task A and re-evolving a new

population of paths for task B, allows task B to be learned faster than it

could be learned from scratch or after fine-tuning. Paths evolved on task B

re-use parts of the optimal path evolved on task A. Positive transfer was

demonstrated for binary MNIST, CIFAR, and SVHN supervised learning

classification tasks, and a set of Atari and Labyrinth reinforcement learning

tasks, suggesting PathNets have general applicability for neural network

training. Finally, PathNet also significantly improves the robustness to

hyperparameter choices of a parallel asynchronous reinforcement learning

algorithm (A3C).

Does Weather Matter? Causal Analysis of TV Logs

Shi Zong , Branislav Kveton , Shlomo Berkovsky , Azin Ashkan , Nikos Vlassis , Zheng Wen Subjects : Computers and Society (cs.CY) ; Learning (cs.LG)

Weather affects our mood and behaviors, and many aspects of our life. When it

is sunny, most people become happier; but when it rains, some people get

depressed. Despite this evidence and the abundance of data, weather has mostly

been overlooked in the machine learning and data science research. This work

presents a causal analysis of how weather affects TV watching patterns. We show

that some weather attributes, such as pressure and precipitation, cause major

changes in TV watching patterns. To the best of our knowledge, this is the

first large-scale causal study of the impact of weather on TV watching

patterns.

A Comparative Study on Different Types of Approaches to Bengali document Categorization

Document categorization is a technique where the category of a document is

determined. In this paper three well-known supervised learning techniques which

are Support Vector Machine(SVM), Na”ive Bayes(NB) and Stochastic Gradient

Descent(SGD) compared for Bengali document categorization. Besides classifier,

classification also depends on how feature is selected from dataset. For

analyzing those classifier performances on predicting a document against twelve

categories several feature selection techniques are also applied in this

article namely Chi square distribution, normalized TFIDF (term

frequency-inverse document frequency) with word analyzer. So, we attempt to

explore the efficiency of those three-classification algorithms by using two

different feature selection techniques in this article.

Self-Adaptation of Activity Recognition Systems to New Sensors

Traditional activity recognition systems work on the basis of training,

taking a fixed set of sensors into account. In this article, we focus on the

question how pattern recognition can leverage new information sources without

any, or with minimal user input. Thus, we present an approach for opportunistic

activity recognition, where ubiquitous sensors lead to dynamically changing

input spaces. Our method is a variation of well-established principles of

machine learning, relying on unsupervised clustering to discover structure in

data and inferring cluster labels from a small number of labeled dates in a

semi-supervised manner. Elaborating the challenges, evaluations of over 3000

sensor combinations from three multi-user experiments are presented in detail

and show the potential benefit of our approach.

Predicting SMT Solver Performance for Software Verification

The Why3 IDE and verification system facilitates the use of a wide range of

Satisfiability Modulo Theories (SMT) solvers through a driver-based

architecture. We present Where4: a portfolio-based approach to discharge Why3

proof obligations. We use data analysis and machine learning techniques on

static metrics derived from program source code. Our approach benefits software

engineers by providing a single utility to delegate proof obligations to the

solvers most likely to return a useful result. It does this in a time-efficient

way using existing Why3 and solver installations – without requiring low-level

knowledge about SMT solver operation from the user.

One Size Fits All : Effectiveness of Local Search on Structured Data

Vincent Cohen-Addad , Chris Schwiegelshohn Subjects : Data Structures and Algorithms (cs.DS) ; Computational Geometry (cs.CG); Learning (cs.LG)

In this paper, we analyze the performance of a simple and standard Local

Search algorithm for clustering on well behaved data. Since the seminal paper

by Ostrovsky, Rabani, Schulman and Swamy [FOCS 2006], much progress has been

made to characterize real-world instances. We distinguish the three main

definitions — Distribution Stability (Awasthi, Blum, Sheffet, FOCS 2010) —

Spectral Separability (Kumar, Kannan, FOCS 2010) — Perturbation Resilience

(Bilu, Linial, ICS 2010) We show that Local Search performs well on the

instances with the aforementioned stability properties. Specifically, for the

(k)-means and (k)-median objective, we show that Local Search exactly recovers

the optimal clustering if the dataset is (3+varepsilon)-perturbation

resilient, and is a PTAS for distribution stability and spectral separability.

This implies the first PTAS for instances satisfying the spectral separability

condition. For the distribution stability condition we also go beyond previous

work by showing that the clustering output by the algorithm and the optimal

clustering are very similar. This is a significant step toward understanding

the success of Local Search heuristics in clustering applications and supports

the legitimacy of the stability conditions: They characterize some of the

structure of real-world instances that make Local Search a popular heuristic.

Feature base fusion for splicing forgery detection based on neuro fuzzy

Habib Ghaffari Hadigheh , Ghazali bin sulong Subjects : Computer Vision and Pattern Recognition (cs.CV) ; Artificial Intelligence (cs.AI); Learning (cs.LG)

Most of researches on image forensics have been mainly focused on detection

of artifacts introduced by a single processing tool. They lead in the

development of many specialized algorithms looking for one or more particular

footprints under specific settings. Naturally, the performance of such

algorithms are not perfect, and accordingly the provided output might be noisy,

inaccurate and only partially correct. Furthermore, a forged image in practical

scenarios is often the result of utilizing several tools available by

image-processing software systems. Therefore, reliable tamper detection

requires developing more poweful tools to deal with various tempering

scenarios. Fusion of forgery detection tools based on Fuzzy Inference System

has been used before for addressing this problem. Adjusting the membership

functions and defining proper fuzzy rules for attaining to better results are

time-consuming processes. This can be accounted as main disadvantage of fuzzy

inference systems. In this paper, a Neuro-Fuzzy inference system for fusion of

forgery detection tools is developed. The neural network characteristic of

these systems provides appropriate tool for automatically adjusting the

membership functions. Moreover, initial fuzzy inference system is generated

based on fuzzy clustering techniques. The proposed framework is implemented and

validated on a benchmark image splicing data set in which three forgery

detection tools are fused based on adaptive Neuro-Fuzzy inference system. The

outcome of the proposed method reveals that applying Neuro Fuzzy inference

systems could be a better approach for fusion of forgery detection tools.

Deep Recurrent Neural Network for Protein Function Prediction from Sequence

Xueliang Liu Subjects : Quantitative Methods (q-bio.QM) ; Learning (cs.LG); Biomolecules (q-bio.BM); Machine Learning (stat.ML)

As high-throughput biological sequencing becomes faster and cheaper, the need

to extract useful information from sequencing becomes ever more paramount,

often limited by low-throughput experimental characterizations. For proteins,

accurate prediction of their functions directly from their primary amino-acid

sequences has been a long standing challenge. Here, machine learning using

artificial recurrent neural networks (RNN) was applied towards classification

of protein function directly from primary sequence without sequence alignment,

heuristic scoring or feature engineering. The RNN models containing

long-short-term-memory (LSTM) units trained on public, annotated datasets from

UniProt achieved high performance for in-class prediction of four important

protein functions tested, particularly compared to other machine learning

algorithms using sequence-derived protein features. RNN models were used also

for out-of-class predictions of phylogenetically distinct protein families with

similar functions, including proteins of the CRISPR-associated nuclease,

ferritin-like iron storage and cytochrome P450 families. Applying the trained

RNN models on the partially unannotated UniRef100 database predicted not only

candidates validated by existing annotations but also currently unannotated

sequences. Some RNN predictions for the ferritin-like iron sequestering

function were experimentally validated, even though their sequences differ

significantly from known, characterized proteins and from each other and cannot

be easily predicted using popular bioinformatics methods. As sequencing and

experimental characterization data increases rapidly, the machine-learning

approach based on RNN could be useful for discovery and prediction of

homologues for a wide range of protein functions.

Information Theory

On the Lattice of Cyclic Linear Codes Over Finite Chain Rings

Alexandre Fotue Tabue , Christophe Mouaha Subjects : Information Theory (cs.IT)

Let ( exttt{R}) be a commutative finite chain ring of invariants ((q,s).) In

this paper, the trace representation of any free cyclic ( exttt{R})-linear

code of length (ell,) is presented, via the (q)-cyclotomic cosets modulo

(ell,) when ( exttt{gcd}(ell, q) = 1.) The lattice

(left( exttt{Cy}( exttt{R},ell), +, cap

ight)) of cyclic

( exttt{R})-linear codes of length (ell,) is investigated. A lower bound on

the Hamming distance of cyclic ( exttt{R})-linear codes of length (ell,) is

established. When (q) is even, a family of MDS and self-orthogonal

( exttt{R})-linear cyclic codes, is constructed.

Contraction of Cyclic Codes Over Finite Chain Rings

Alexandre Fotue Tabue , Christophe Mouaha Subjects : Information Theory (cs.IT)

Let ( exttt{R}) be a commutative finite chain ring of invariants ((q,s)) and

(Gamma( exttt{R})) the Teichm”uller’s set of ( exttt{R}.) In this paper,

the trace representation cyclic ( exttt{R})-linear codes of length (ell,) is

presented, when ( exttt{gcd}(ell, q) = 1.) We will show that the contractions

of some cyclic ( exttt{R})-linear codes of length (uell) are

(gamma)-constacyclic ( exttt{R})-linear codes of length (ell,) where

(gammainGamma( exttt{R})) and the multiplicative order of is (u.)

On the Computation of the Shannon Capacity of a Discrete Channel with Noise

Muroga [M52] showed how to express the Shannon channel capacity of a discrete

channel with noise [S49] as an explicit function of the transition

probabilities. His method accommodates channels with any finite number of input

symbols, any finite number of output symbols and any transition probability

matrix. Silverman [S55] carried out Muroga’s method in the special case of a

binary channel (and went on to analyse “cascades” of several such binary

channels).

This article is a note on the resulting formula for the capacity C(a, c) of a

single binary channel. We aim to clarify some of the arguments and correct a

small error. In service of this aim, we first formulate several of Shannon’s

definitions and proofs in terms of discrete measure-theoretic probability

theory. We provide an alternate proof to Silverman’s, of the feasibility of the

optimal input distribution for a binary channel. For convenience, we also

express C(a, c) in a single expression explicitly dependent on a and c only,

which Silverman stopped short of doing.

Variable-Length Resolvability for General Sources and Channels

We introduce the problem of variable-length source resolvability, where a

given target probability distribution is approximated by encoding a

variable-length uniform random number, and the asymptotically minimum average

length rate of the uniform random numbers, called the (variable-length)

resolvability, is investigated. We first analyze the variable-length

resolvability with the variational distance as an approximation measure. Next,

we investigate the case under the divergence as an approximation measure. When

the asymptotically exact approximation is required, it is shown that the

resolvability under the two kinds of approximation measures coincides. We then

extend the analysis to the case of channel resolvability, where the target

distribution is the output distribution via a general channel due to the fixed

general source as an input. The obtained characterization of the channel

resolvability is fully general in the sense that when the channel is just the

identity mapping, the characterization reduces to the general formula for the

source resolvability. We also analyze the second-order variable-length

resolvability.

Signal Recovery from Unlabeled Samples

In this paper, we study the recovery of a signal from a collection of

unlabeled and possibly noisy measurements via a measurement matrix with random

i.i.d. Gaussian components. We call the measurements unlabeled since their

order is missing, namely, it is not known a priori which elements of the

resulting measurements correspond to which row of the measurement matrix. We

focus on the special case of ordered measurements, where only a subset of the

measurements is kept and the order of the taken measurements is preserved. We

identify a natural duality between this problem and the traditional Compressed

Sensing, where we show that the unknown support (location of nonzero elements)

of a sparse signal in Compressed Sensing corresponds in a natural way to the

unknown location of the measurements kept in unlabeled sensing. While in

Compressed Sensing it is possible to recover a sparse signal from an

under-determined set of linear equations (less equations than the dimension of

the signal), successful recovery in unlabeled sensing requires taking more

samples than the dimension of the signal. We develop a low-complexity

alternating minimization algorithm to recover the initial signal from the set

of its unlabeled samples. We also study the behavior of the proposed algorithm

for different signal dimensions and number of measurements both theoretically

and empirically via numerical simulations. The results are a reminiscent of the

phase-transition similar to that occurring in Compressed Sensing.

Ultra Reliable Communication via Optimum Power Allocation for Type-I ARQ in Finite Block-Length

We analyze the performance of the type-I automatic repeat request (ARQ)

protocol with ultra-reliability constraints. First, we show that achieving a

very low packet outage probability by using an open loop setup is a difficult

task. Thus, we introduce the ARQ protocol as a solution for achieving the

required low outage probabilities for ultra reliable communication. For this

protocol, we present an optimal power allocation scheme that would allow us to

reach any outage probability target in the finite block-length regime. We

formulate the power allocation problem as minimization of the average

transmitted power under a given outage probability and maximum transmit power

constraint. By utilizing the Karush-Kuhn-Tucker (KKT) conditions, we solve the

optimal power allocation problem and provide a closed form solution. Next, we

analyze the effect of implementing the ARQ protocol on the throughput. We show

that by using the proposed power allocation scheme we can minimize the loss of

throughput that is caused from the retransmissions. Furthermore, we analyze the

effect of the feedback delay length in our scheme.

Low Dimensional Atomic Norm Representations in Line Spectral Estimation

Maxime Ferreira Da Costa , Wei Dai Subjects : Information Theory (cs.IT)

The line spectral estimation problem consists in recovering the frequencies

of a complex valued time signal that is assumed to be sparse in the spectral

domain from its discrete observations. Unlike the gridding required by the

classical compressed sensing framework, line spectral estimation reconstructs

signals whose spectral supports lie continuously in the Fourier domain. If

recent advances have shown that atomic norm relaxation produces highly robust

estimates in this context, the computational cost of this approach remains,

however, the major flaw for its application to practical systems.

In this work, we aim to bridge the complexity issue by studying the atomic

norm minimization problem from low dimensional projection of the signal

samples. We derive conditions on the sub-sampling matrix under which the

partial atomic norm can be expressed by a low-dimensional semidefinite program.

Moreover, we illustrate the tightness of this relaxation by showing that it is

possible to recover the original signal in poly-logarithmic time for two

specific sub-sampling patterns.

Optimal Transport to the Entropy-Power Inequality and a Reverse Inequality

Olivier Rioul Subjects : Information Theory (cs.IT)

We present a simple proof of the entropy-power inequality using an optimal

transportation argument which takes the form of a simple change of variables.

The same argument yields a reverse inequality involving a conditional

differential entropy which has its own interest. For each inequality, the

equality case is easily captured by this method and the proof is formally

identical in one and several dimensions.

Non-Orthogonal Multiple Access Schemes in Wireless Powered Communication Networks

We characterize time and power allocations to optimize the sum-throughput of

a Wireless Powered Communication Network (WPCN) with Non-Orthogonal Multiple

Access (NOMA). In our setup, an Energy Rich (ER) source broadcasts wireless

energy to several devices, which use it to simultaneously transmit data to an

Access Point (AP) on the uplink. Differently from most prior works, in this

paper we consider a generic scenario, in which the ER and AP do not coincide,

i.e., two separate entities. We study two NOMA decoding schemes, namely Low

Complexity Decoding (LCD) and Successive Interference Cancellation Decoding

(SICD). For each scheme, we formulate a sum-throughput optimization problem

over a finite horizon. Despite the complexity of the LCD optimization problem,

attributed to its non-convexity, we recast it into a series of geometric

programs. On the other hand, we establish the convexity of the SICD

optimization problem and propose an algorithm to find its optimal solution. Our

numerical results demonstrate the importance of using successive interference

cancellation in WPCNs with NOMA, and show how the energy should be distributed

as a function of the system parameters.

Fast and Lightweight Rate Control for Onboard Predictive Coding of Hyperspectral Images

Diego Valsesia , Enrico Magli Subjects : Information Theory (cs.IT)

Predictive coding is attractive for compression of hyperspecral images

onboard of spacecrafts in light of the excellent rate-distortion performance

and low complexity of recent schemes. In this letter we propose a rate control

algorithm and integrate it in a lossy extension to the CCSDS-123 lossless

compression recommendation. The proposed rate algorithm overhauls our previous

scheme by being orders of magnitude faster and simpler to implement, while

still providing the same accuracy in terms of output rate and comparable or

better image quality.

On Zero Error Capacity of Nearest Neighbor Error Channels with Multilevel Alphabet

Takafumi Nakano , Tadashi Wadayama Subjects : Information Theory (cs.IT)

This paper studies the zero error capacity of the Nearest Neighbor Error

(NNE) channels with a multilevel alphabet. In the NNE channels, a transmitted

symbol is a (d)-tuple of elements in ({0,1,2,dots, n-1 }). It is assumed

that only one element error to a nearest neighbor element in a transmitted

symbol can occur. The NNE channels can be considered as a special type of

limited magnitude error channels, and it is closely related to error models for

flash memories. In this paper, we derive a lower bound of the zero error

capacity of the NNE channels based on a result of the perfect Lee codes. An

upper bound of the zero error capacity of the NNE channels is also derived from

a feasible solution of a linear programming problem defined based on the

confusion graphs of the NNE channels. As a result, a concise formula of the

zero error capacity is obtained using the lower and upper bounds.

Communication Cost of Transforming a Nearest Plane Partition to the Voronoi Partition

We consider the problem of distributed computation of the nearest lattice

point for a two dimensional lattice. An interactive model of communication is

considered. We address the problem of reconfiguring a specific rectangular

partition, a nearest plane, or Babai, partition, into the Voronoi partition.

Expressions are derived for the error probability as a function of the total

number of communicated bits. With an infinite number of allowed communication

rounds, the average cost of achieving zero error probability is shown to be

finite. For the interactive model, with a single round of communication,

expressions are obtained for the error probability as a function of the bits

exchanged. We observe that the error exponent depends on the lattice.

On the Communication Cost of Determining an Approximate Nearest Lattice Point

We consider the closest lattice point problem in a distributed network

setting and study the communication cost and the error probability for

computing an approximate nearest lattice point, using the nearest-plane

algorithm, due to Babai. Two distinct communication models, centralized and

interactive, are considered. The importance of proper basis selection is

addressed. Assuming a reduced basis for a two-dimensional lattice, we determine

the approximation error of the nearest plane algorithm. The communication cost

for determining the Babai point, or equivalently, for constructing the

rectangular nearest-plane partition, is calculated in the interactive setting.

For the centralized model, an algorithm is presented for reducing the

communication cost of the nearest plane algorithm in an arbitrary number of

dimensions.

Steady-state performance analysis of the recursive maximum correntropy algorithm and its application in adaptive beamforming with alpha-stable noise

Lu Lu , Haiquan Zhao Subjects : Information Theory (cs.IT)

As a well-established adaptation criterion, the maximum correntropy criterion

(MCC) has received increased attention due to its robustness against outliers.

In this paper, a new complex recursive maximum correntropy (CRMC) algorithm

without any priori information on the noise characteristics, is proposed under

the MCC. We first study the steady-state excess mean-square-error (EMSE)

behavior of the CRMC algorithm by using energy conservation relation and some

reasonable approximations. Then, the proposed algorithm is introduced to

adaptive beamforming problem, where the desired signal is contaminated by the

impulsive noises. The results obtained from simulation study establish the

effectiveness of this new beamformer.

Integer-Forcing Message Recovering in Interference Channels

In this paper, we propose a scheme referred to as integer-forcing message

recovering (IFMR) to enable receivers to recover their desirable messages in

interference channels. Compared to the state-of-the- art integer-forcing linear

receiver (IFLR), our proposed IFMR approach needs to decode considerably less

number of messages. In our method, each receiver recovers independent linear

integer combinations of the desirable messages each from two independent

equations. We propose an efficient algorithm to sequentially find the equations

and integer combinations with maximum rates. We evaluate the performance of our

scheme and compare the results with the minimum mean-square error (MMSE) and

zero-forcing (ZF), as well as the IFLR schemes. The results indicate that our

IFMR scheme outperforms the MMSE and ZF schemes, in terms of achievable rate,

considerably. Also, compared to IFLR, the IFMR scheme achieves slightly less

rates in moderate signal-to-noise ratios, with significantly less

implementation complexity.

Channel Resolvability Theorems for General Sources and Channels

In the problem of channel resolvability, where a given output probability

distribution via a channel is approximated by transforming the uniform random

numbers, characterizing the asymptotically minimum rate of the size of the

random numbers, called the channel resolvability, has been open. This paper

derives formulas for the channel resolvability for a given general source and

channel pair. We also investigate the channel resolvability in an optimistic

sense. It is demonstrated that the derived general formulas recapture a

single-letter formula for the stationary memoryless source and channel. When

the channel is the identity mapping, the established formulas reduce to an

alternative form of the spectral sup-entropy rates, which play a key role in

information spectrum methods. The analysis is also extended to the second-order

channel resolvability.

Scheduling Status Updates to Minimize Age of Information with an Energy Harvesting Sensor

Age of Information is a measure of the freshness of status updates in

monitoring applications and update-based systems. We study a real-time remote

sensing scenario with a sensor which is restricted by time-varying energy

constraints and battery limitations. The sensor sends updates over a packet

erasure channel with no feedback. The problem of finding an age-optimal

threshold policy, with the transmission threshold being a function of the

energy state and the estimated current age, is formulated. The average age is

analyzed for the unit battery scenario under a memoryless energy arrival

process. Somewhat surprisingly, for any finite arrival rate of energy, there is

a positive age threshold for transmission, which corresponding to transmitting

at a rate lower than that dictated by the rate of energy arrivals. A lower

bound on the average age is obtained for general battery size.

Construction of Fixed Rate Non-Binary WOM Codes based on Integer Programming

Yoju Fujino , Tadashi Wadayama Subjects : Information Theory (cs.IT)

In this paper, we propose a construction of non-binary WOM

(Write-Once-Memory) codes for WOM storages such as flash memories. The WOM

codes discussed in this paper are fixed rate WOM codes where messages in a

fixed alphabet of size (M) can be sequentially written in the WOM storage at

least (t^*)-times. In this paper, a WOM storage is modeled by a state

transition graph. The proposed construction has the following two features.

First, it includes a systematic method to determine the encoding regions in the

state transition graph. Second, the proposed construction includes a labeling

method for states by using integer programming. Several novel WOM codes for (q)

level flash memories with 2 cells are constructed by the proposed construction.

They achieve the worst numbers of writes (t^*) that meet the known upper bound

in many cases. In addition, we constructed fixed rate non-binary WOM codes with

the capability to reduce ICI (inter cell interference) of flash cells. One of

the advantages of the proposed construction is its flexibility. It can be

applied to various storage devices, to various dimensions (i.e, number of

cells), and various kind of additional constraints.

On Cooperation and Interference in the Weak Interference Regime (Full Version with Detailed Proofs)

Daniel Zahavi , Ron Dabora Subjects : Information Theory (cs.IT)

Handling interference is one of the main challenges in the design of wireless

networks. In this paper we study the application of cooperation for

interference management in the weak interference (WI) regime, focusing on the

Z-interference channel with a causal relay (Z-ICR), when the channel

coefficients are subject to ergodic phase fading, all transmission powers are

finite, and the relay is full-duplex. In order to provide a comprehensive

understanding of the benefits of cooperation in the WI regime, we characterize,

for the first time, two major performance measures for the ergodic phase fading

Z-ICR in the WI regime: The sum-rate capacity and the maximal generalized

degrees-of-freedom (GDoF). In the capacity analysis, we obtain conditions on

the channel coefficients, subject to which the sum-rate capacity of the ergodic

phase fading Z-ICR is achieved by treating interference as noise at each

receiver, and explicitly state the corresponding sum-rate capacity. In the GDoF

analysis, we derive conditions on the exponents of the magnitudes of the

channel coefficients, under which treating interference as noise achieves the

maximal GDoF, which is explicitly characterized as well. It is shown that under

certain conditions on the channel coefficients, {em relaying strictly

increases} both the sum-rate capacity and the maximal GDoF of the ergodic phase

fading Z-interference channel in the WI regime. Our results demonstrate {em

for the first time} the gains from relaying in the presence of interference,

{em when interference is weak and the relay power is finite}, both in

increasing the sum-rate capacity and in increasing the maximal GDoF, compared

to the channel without a relay.

Multilevel Code Construction for Compound Fading Channels

We consider explicit constructions of multi-level lattice codes that

universally approach the capacity of the compound block-fading channel.

Specifically, building on algebraic partitions of lattices, we show how to

construct codes with negligible probability of error for any channel

realization and normalized log-density approaching the Poltyrev limit. Capacity

analyses and numerical results on the achievable rates for each partition level

are provided. The proposed codes have several enjoyable properties such as

constructiveness and good decoding complexity, as compared to random one-level

codes. Numerical results for finite-dimensional multi-level lattices based on

polar codes are exhibited.

On the Fronthaul Statistical Multiplexing Gain

Breaking the fronthaul capacity limitations is vital to make cloud radio

access network (C-RAN) scalable and practical. One promising way is aggregating

several remote radio units (RRUs) as a cluster to share a fronthaul link, so as

to enjoy the statistical multiplexing gain brought by the spatial randomness of

the traffic. In this letter, a tractable model is proposed to analyze the

fronthaul statistical multiplexing gain. We first derive the user blocking

probability caused by the limited fronthaul capacity, including its upper and

lower bounds. We then obtain the limits of fronthaul statistical multiplexing

gain when the cluster size approaches infinity. Analytical results reveal that

the user blocking probability decreases exponentially with the average

fronthaul capacity per RRU, and the exponent is proportional to the cluster

size. Numerical results further show considerable fronthaul statistical

multiplexing gain even at a small to medium cluster size.

Entropic Causality and Greedy Minimum Entropy Coupling

We study the problem of identifying the causal relationship between two

discrete random variables from observational data. We recently proposed a novel

framework called entropic causality that works in a very general functional

model but makes the assumption that the unobserved exogenous variable has small

entropy in the true causal direction.

This framework requires the solution of a minimum entropy coupling problem:

Given marginal distributions of m discrete random variables, each on n states,

find the joint distribution with minimum entropy, that respects the given

marginals. This corresponds to minimizing a concave function of nm variables

over a convex polytope defined by nm linear constraints, called a

transportation polytope. Unfortunately, it was recently shown that this minimum

entropy coupling problem is NP-hard, even for 2 variables with n states. Even

representing points (joint distributions) over this space can require

exponential complexity (in n, m) if done naively.

In our recent work we introduced an efficient greedy algorithm to find an

approximate solution for this problem. In this paper we analyze this algorithm

and establish two results: that our algorithm always finds a local minimum and

also is within an additive approximation error from the unknown global optimum.

Sampling Without Time: Recovering Echoes of Light via Temporal Phase Retrieval

This paper considers the problem of sampling and reconstruction of a

continuous-time sparse signal without assuming the knowledge of the sampling

instants or the sampling rate. This topic has its roots in the problem of

recovering multiple echoes of light from its low-pass filtered and

auto-correlated, time-domain measurements. Our work is closely related to the

topic of sparse phase retrieval and in this context, we discuss the advantage

of phase-free measurements. While this problem is ill-posed, cues based on

physical constraints allow for its appropriate regularization. We validate our

theory with experiments based on customized, optical time-of-flight imaging

sensors. What singles out our approach is that our sensing method allows for

temporal phase retrieval as opposed to the usual case of spatial phase

retrieval. Preliminary experiments and results demonstrate a compelling

capability of our phase-retrieval based imaging device.

Rotated Eigenstructure Analysis for Source Localization without Energy-decay Models

Junting Chen , Urbashi Mitra Subjects : Information Theory (cs.IT)

Herein, the problem of simultaneous localization of two sources given a

modest number of samples is examined. In particular, the strategy does not

require knowledge of the target signatures of the sources a priori, nor does it

exploit classical methods based on a particular decay rate of the energy

emitted from the sources as a function of range. General structural properties

of the signatures such as unimodality are exploited. The algorithm localizes

targets based on the rotated eigenstructure of a reconstructed observation

matrix. In particular, the optimal rotation can be found by maximizing the

ratio of the dominant singular value of the observation matrix over the nuclear

norm of the optimally rotated observation matrix. It is shown that this ratio

has a unique local maximum leading to computationally efficient search

algorithms. Moreover, analytical results are developed to show that the squared

localization error decreases at a rate faster than the baseline scheme.

An asymptotic equipartition property for measures on model spaces

Let (G) be a sofic group, and let (Sigma = (sigma_n)_{ngeq 1}) be a sofic

approximation to it. For a probability-preserving (G)-system, a variant of the

sofic entropy relative to (Sigma) has recently been defined in terms of

sequences of measures on its model spaces that `converge’ to the system in a

certain sense. Here we prove that, in order to study this notion, one may

restrict attention to those sequences that have the asymptotic equipartition

property. This may be seen as a relative in the sofic setting of the

Shannon–McMillan theorem.

We also give some first applications of this result, including a new formula

for the sofic entropy of a ((G imes H))-system obtained by co-induction from a

(G)-system, where (H) is any other infinite sofic group.

Pure Rough Mereology and Counting

The study of mereology (parts and wholes) in the context of formal approaches

to vagueness can be approached in a number of ways. In the context of rough

sets, mereological concepts with a set-theoretic or valuation based ontology

acquire complex and diverse behavior. In this research a general rough set

framework called granular operator spaces is extended and the nature of

parthood in it is explored from a minimally intrusive point of view. This is

used to develop counting strategies that help in classifying the framework. The

developed methodologies would be useful for drawing involved conclusions about

the nature of data (and validity of assumptions about it) from antichains

derived from context. The problem addressed is also about whether counting

procedures help in confirming that the approximations involved in formation of

data are indeed rough approximations?

Low Rank Magnetic Resonance Fingerprinting

Magnetic Resonance Fingerprinting (MRF) is a relatively new approach that

provides quantitative MRI measures using randomized acquisition. Extraction of

physical quantitative tissue parameters is performed off-line, based on

acquisition with varying parameters and a dictionary generated according to the

Bloch equations. MRF uses hundreds of radio frequency (RF) excitation pulses

for acquisition, and therefore high under-sampling ratio in the sampling domain

(k-space) is required for reasonable scanning time. This under-sampling causes

spatial artifacts that hamper the ability to accurately estimate the tissue’s

quantitative values. In this work, we introduce a new approach for quantitative

MRI using MRF, called magnetic resonance Fingerprinting with LOw Rank (FLOR).

We exploit the low rank property of the concatenated temporal imaging

contrasts, on top of the fact that the MRF signal is sparsely represented in

the generated dictionary domain. We present an iterative scheme that consists

of a gradient step followed by a low rank projection using the singular value

decomposition. Experiments on real MRI data, acquired using a spirally-sampled

MRF FISP sequence, demonstrate improved resolution compared to other

compressed-sensing based methods for MRF at 5% sampling ratio.