July 29, 2019

3160 words 15 mins read

Paper Group ANR 139

Building Detection from Satellite Images on a Global Scale. Spintronics based Stochastic Computing for Efficient Bayesian Inference System. A Novel Deep Learning Architecture for Testis Histology Image Classification. Grouped Convolutional Neural Networks for Multivariate Time Series. Class Specific Feature Selection for Interval Valued Data Throug …

Building Detection from Satellite Images on a Global Scale


Title	Building Detection from Satellite Images on a Global Scale
Authors	Amy Zhang, Xianming Liu, Andreas Gros, Tobias Tiecke
Abstract	In the last several years, remote sensing technology has opened up the possibility of performing large scale building detection from satellite imagery. Our work is some of the first to create population density maps from building detection on a large scale. The scale of our work on population density estimation via high resolution satellite images raises many issues, that we will address in this paper. The first was data acquisition. Labeling buildings from satellite images is a hard problem, one where we found our labelers to only be about 85% accurate at. There is a tradeoff of quantity vs. quality of labels, so we designed two separate policies for labels meant for training sets and those meant for test sets, since our requirements of the two set types are quite different. We also trained weakly supervised footprint detection models with the classification labels, and semi-supervised approaches with a small number of pixel-level labels, which are very expensive to procure.
Tasks	Density Estimation
Published	2017-07-27
URL	http://arxiv.org/abs/1707.08952v1
PDF	http://arxiv.org/pdf/1707.08952v1.pdf
PWC	https://paperswithcode.com/paper/building-detection-from-satellite-images-on-a
Repo
Framework

Spintronics based Stochastic Computing for Efficient Bayesian Inference System


Title	Spintronics based Stochastic Computing for Efficient Bayesian Inference System
Authors	Xiaotao Jia, Jianlei Yang, Zhaohao Wang, Yiran Chen, Hai, Li, Weisheng Zhao
Abstract	Bayesian inference is an effective approach for solving statistical learning problems especially with uncertainty and incompleteness. However, inference efficiencies are physically limited by the bottlenecks of conventional computing platforms. In this paper, an emerging Bayesian inference system is proposed by exploiting spintronics based stochastic computing. A stochastic bitstream generator is realized as the kernel components by leveraging the inherent randomness of spintronics devices. The proposed system is evaluated by typical applications of data fusion and Bayesian belief networks. Simulation results indicate that the proposed approach could achieve significant improvement on inference efficiencies in terms of power consumption and inference speed.
Tasks	Bayesian Inference
Published	2017-11-03
URL	http://arxiv.org/abs/1711.01125v1
PDF	http://arxiv.org/pdf/1711.01125v1.pdf
PWC	https://paperswithcode.com/paper/spintronics-based-stochastic-computing-for
Repo
Framework

A Novel Deep Learning Architecture for Testis Histology Image Classification


Title	A Novel Deep Learning Architecture for Testis Histology Image Classification
Authors	Chia-Yu Kao, Leonard McMillan
Abstract	Unlike other histology analysis, classification of tubule status in testis histology is very challenging due to their high similarity of texture and shape. Traditional deep learning networks have difficulties to capture nuance details among different tubule categories. In this paper, we propose a novel deep learning architecture for feature learning, image classification, and image reconstruction. It is based on stacked auto-encoders with an additional layer, called a hyperlayer, which is created to capture features of an image at different layers in the network. This addition effectively combines features at different scales and thus provides a more complete profile for further classification. Evaluation is performed on a set of 10,542 tubule image patches. We demonstrate our approach with two experiments on two different subsets of the dataset. The results show that the features learned from our architecture achieve more than 98% accuracy and represent an improvement over traditional deep network architectures.
Tasks	Image Classification, Image Reconstruction
Published	2017-07-18
URL	http://arxiv.org/abs/1707.05809v1
PDF	http://arxiv.org/pdf/1707.05809v1.pdf
PWC	https://paperswithcode.com/paper/a-novel-deep-learning-architecture-for-testis
Repo
Framework

Grouped Convolutional Neural Networks for Multivariate Time Series


Title	Grouped Convolutional Neural Networks for Multivariate Time Series
Authors	Subin Yi, Janghoon Ju, Man-Ki Yoon, Jaesik Choi
Abstract	Analyzing multivariate time series data is important for many applications such as automated control, fault diagnosis and anomaly detection. One of the key challenges is to learn latent features automatically from dynamically changing multivariate input. In visual recognition tasks, convolutional neural networks (CNNs) have been successful to learn generalized feature extractors with shared parameters over the spatial domain. However, when high-dimensional multivariate time series is given, designing an appropriate CNN model structure becomes challenging because the kernels may need to be extended through the full dimension of the input volume. To address this issue, we present two structure learning algorithms for deep CNN models. Our algorithms exploit the covariance structure over multiple time series to partition input volume into groups. The first algorithm learns the group CNN structures explicitly by clustering individual input sequences. The second algorithm learns the group CNN structures implicitly from the error backpropagation. In experiments with two real-world datasets, we demonstrate that our group CNNs outperform existing CNN based regression methods.
Tasks	Anomaly Detection, Time Series
Published	2017-03-29
URL	http://arxiv.org/abs/1703.09938v4
PDF	http://arxiv.org/pdf/1703.09938v4.pdf
PWC	https://paperswithcode.com/paper/grouped-convolutional-neural-networks-for
Repo
Framework

Class Specific Feature Selection for Interval Valued Data Through Interval K-Means Clustering


Title	Class Specific Feature Selection for Interval Valued Data Through Interval K-Means Clustering
Authors	D. S. Guru, N. Vinay Kumar
Abstract	In this paper, a novel feature selection approach for supervised interval valued features is proposed. The proposed approach takes care of selecting the class specific features through interval K-Means clustering. The kernel of K-Means clustering algorithm is modified to adapt interval valued data. During training, a set of samples corresponding to a class is fed into the interval K-Means clustering algorithm, which clusters features into K distinct clusters. Hence, there are K number of features corresponding to each class. Subsequently, corresponding to each class, the cluster representatives are chosen. This procedure is repeated for all the samples of remaining classes. During testing the feature indices correspond to each class are used for validating the given dataset through classification using suitable symbolic classifiers. For experimentation, four standard supervised interval datasets are used. The results show the superiority of the proposed model when compared with the other existing state-of-the-art feature selection methods.
Tasks	Feature Selection
Published	2017-05-31
URL	http://arxiv.org/abs/1705.10986v1
PDF	http://arxiv.org/pdf/1705.10986v1.pdf
PWC	https://paperswithcode.com/paper/class-specific-feature-selection-for-interval
Repo
Framework

On The Robustness of a Neural Network


Title	On The Robustness of a Neural Network
Authors	El Mahdi El Mhamdi, Rachid Guerraoui, Sebastien Rouault
Abstract	With the development of neural networks based machine learning and their usage in mission critical applications, voices are rising against the \textit{black box} aspect of neural networks as it becomes crucial to understand their limits and capabilities. With the rise of neuromorphic hardware, it is even more critical to understand how a neural network, as a distributed system, tolerates the failures of its computing nodes, neurons, and its communication channels, synapses. Experimentally assessing the robustness of neural networks involves the quixotic venture of testing all the possible failures, on all the possible inputs, which ultimately hits a combinatorial explosion for the first, and the impossibility to gather all the possible inputs for the second. In this paper, we prove an upper bound on the expected error of the output when a subset of neurons crashes. This bound involves dependencies on the network parameters that can be seen as being too pessimistic in the average case. It involves a polynomial dependency on the Lipschitz coefficient of the neurons activation function, and an exponential dependency on the depth of the layer where a failure occurs. We back up our theoretical results with experiments illustrating the extent to which our prediction matches the dependencies between the network parameters and robustness. Our results show that the robustness of neural networks to the average crash can be estimated without the need to neither test the network on all failure configurations, nor access the training set used to train the network, both of which are practically impossible requirements.
Tasks
Published	2017-07-25
URL	http://arxiv.org/abs/1707.08167v2
PDF	http://arxiv.org/pdf/1707.08167v2.pdf
PWC	https://paperswithcode.com/paper/on-the-robustness-of-a-neural-network
Repo
Framework

An efficient genetic algorithm for large-scale planning of robust industrial wireless networks


Title	An efficient genetic algorithm for large-scale planning of robust industrial wireless networks
Authors	Xu Gong, David Plets, Emmeric Tanghe, Toon De Pessemier, Luc Martens, Wout Joseph
Abstract	An industrial indoor environment is harsh for wireless communications compared to an office environment, because the prevalent metal easily causes shadowing effects and affects the availability of an industrial wireless local area network (IWLAN). On the one hand, it is costly, time-consuming, and ineffective to perform trial-and-error manual deployment of wireless nodes. On the other hand, the existing wireless planning tools only focus on office environments such that it is hard to plan IWLANs due to the larger problem size and the deployed IWLANs are vulnerable to prevalent shadowing effects in harsh industrial indoor environments. To fill this gap, this paper proposes an overdimensioning model and a genetic algorithm based over-dimensioning (GAOD) algorithm for deploying large-scale robust IWLANs. As a progress beyond the state-of-the-art wireless planning, two full coverage layers are created. The second coverage layer serves as redundancy in case of shadowing. Meanwhile, the deployment cost is reduced by minimizing the number of access points (APs); the hard constraint of minimal inter-AP spatial paration avoids multiple APs covering the same area to be simultaneously shadowed by the same obstacle. The computation time and occupied memory are dedicatedly considered in the design of GAOD for large-scale optimization. A greedy heuristic based over-dimensioning (GHOD) algorithm and a random OD algorithm are taken as benchmarks. In two vehicle manufacturers with a small and large indoor environment, GAOD outperformed GHOD with up to 20% less APs, while GHOD outputted up to 25% less APs than a random OD algorithm. Furthermore, the effectiveness of this model and GAOD was experimentally validated with a real deployment system.
Tasks
Published	2017-08-12
URL	http://arxiv.org/abs/1709.04321v1
PDF	http://arxiv.org/pdf/1709.04321v1.pdf
PWC	https://paperswithcode.com/paper/an-efficient-genetic-algorithm-for-large-1
Repo
Framework

Understanding Aesthetics in Photography using Deep Convolutional Neural Networks


Title	Understanding Aesthetics in Photography using Deep Convolutional Neural Networks
Authors	Maciej Suchecki, Tomasz Trzcinski
Abstract	Evaluating aesthetic value of digital photographs is a challenging task, mainly due to numerous factors that need to be taken into account and subjective manner of this process. In this paper, we propose to approach this problem using deep convolutional neural networks. Using a dataset of over 1.7 million photos collected from Flickr, we train and evaluate a deep learning model whose goal is to classify input images by analysing their aesthetic value. The result of this work is a publicly available Web-based application that can be used in several real-life applications, e.g. to improve the workflow of professional photographers by pre-selecting the best photos.
Tasks
Published	2017-07-27
URL	http://arxiv.org/abs/1707.08985v2
PDF	http://arxiv.org/pdf/1707.08985v2.pdf
PWC	https://paperswithcode.com/paper/understanding-aesthetics-in-photography-using
Repo
Framework

Supervised Hashing with End-to-End Binary Deep Neural Network


Title	Supervised Hashing with End-to-End Binary Deep Neural Network
Authors	Dang-Khoa Le Tan, Thanh-Toan Do, Ngai-Man Cheung
Abstract	Image hashing is a popular technique applied to large scale content-based visual retrieval due to its compact and efficient binary codes. Our work proposes a new end-to-end deep network architecture for supervised hashing which directly learns binary codes from input images and maintains good properties over binary codes such as similarity preservation, independence, and balancing. Furthermore, we also propose a new learning scheme that can cope with the binary constrained loss function. The proposed algorithm not only is scalable for learning over large-scale datasets but also outperforms state-of-the-art supervised hashing methods, which are illustrated throughout extensive experiments from various image retrieval benchmarks.
Tasks	Image Retrieval
Published	2017-11-24
URL	http://arxiv.org/abs/1711.08901v2
PDF	http://arxiv.org/pdf/1711.08901v2.pdf
PWC	https://paperswithcode.com/paper/supervised-hashing-with-end-to-end-binary
Repo
Framework

A Fast Noniterative Algorithm for Compressive Sensing Using Binary Measurement Matrices


Title	A Fast Noniterative Algorithm for Compressive Sensing Using Binary Measurement Matrices
Authors	Mahsa Lotfi, Mathukumalli Vidyasagar
Abstract	In this paper we present a new algorithm for compressive sensing that makes use of binary measurement matrices and achieves exact recovery of ultra sparse vectors, in a single pass and without any iterations. Due to its noniterative nature, our algorithm is hundreds of times faster than $\ell_1$-norm minimization, and methods based on expander graphs, both of which require multiple iterations. Our algorithm can accommodate nearly sparse vectors, in which case it recovers index set of the largest components, and can also accommodate burst noise measurements. Compared to compressive sensing methods that are guaranteed to achieve exact recovery of all sparse vectors, our method requires fewer measurements However, methods that achieve statistical recovery, that is, recovery of almost all but not all sparse vectors, can require fewer measurements than our method.
Tasks	Compressive Sensing
Published	2017-08-11
URL	http://arxiv.org/abs/1708.03608v2
PDF	http://arxiv.org/pdf/1708.03608v2.pdf
PWC	https://paperswithcode.com/paper/a-fast-noniterative-algorithm-for-compressive
Repo
Framework

Power Optimizations in MTJ-based Neural Networks through Stochastic Computing


Title	Power Optimizations in MTJ-based Neural Networks through Stochastic Computing
Authors	Ankit Mondal, Ankur Srivastava
Abstract	Artificial Neural Networks (ANNs) have found widespread applications in tasks such as pattern recognition and image classification. However, hardware implementations of ANNs using conventional binary arithmetic units are computationally expensive, energy-intensive and have large area overheads. Stochastic Computing (SC) is an emerging paradigm which replaces these conventional units with simple logic circuits and is particularly suitable for fault-tolerant applications. Spintronic devices, such as Magnetic Tunnel Junctions (MTJs), are capable of replacing CMOS in memory and logic circuits. In this work, we propose an energy-efficient use of MTJs, which exhibit probabilistic switching behavior, as Stochastic Number Generators (SNGs), which forms the basis of our NN implementation in the SC domain. Further, error resilient target applications of NNs allow us to introduce Approximate Computing, a framework wherein accuracy of computations is traded-off for substantial reductions in power consumption. We propose approximating the synaptic weights in our MTJ-based NN implementation, in ways brought about by properties of our MTJ-SNG, to achieve energy-efficiency. We design an algorithm that can perform such approximations within a given error tolerance in a single-layer NN in an optimal way owing to the convexity of the problem formulation. We then use this algorithm and develop a heuristic approach for approximating multi-layer NNs. To give a perspective of the effectiveness of our approach, a 43% reduction in power consumption was obtained with less than 1% accuracy loss on a standard classification problem, with 26% being brought about by the proposed algorithm.
Tasks	Image Classification
Published	2017-08-17
URL	http://arxiv.org/abs/1709.04322v1
PDF	http://arxiv.org/pdf/1709.04322v1.pdf
PWC	https://paperswithcode.com/paper/power-optimizations-in-mtj-based-neural
Repo
Framework

A Hybrid Deep Learning Architecture for Privacy-Preserving Mobile Analytics


Title	A Hybrid Deep Learning Architecture for Privacy-Preserving Mobile Analytics
Authors	Seyed Ali Osia, Ali Shahin Shamsabadi, Sina Sajadmanesh, Ali Taheri, Kleomenis Katevas, Hamid R. Rabiee, Nicholas D. Lane, Hamed Haddadi
Abstract	Internet of Things (IoT) devices and applications are being deployed in our homes and workplaces. These devices often rely on continuous data collection to feed machine learning models. However, this approach introduces several privacy and efficiency challenges, as the service operator can perform unwanted inferences on the available data. Recently, advances in edge processing have paved the way for more efficient, and private, data processing at the source for simple tasks and lighter models, though they remain a challenge for larger, and more complicated models. In this paper, we present a hybrid approach for breaking down large, complex deep neural networks for cooperative, privacy-preserving analytics. To this end, instead of performing the whole operation on the cloud, we let an IoT device to run the initial layers of the neural network, and then send the output to the cloud to feed the remaining layers and produce the final result. In order to ensure that the user’s device contains no extra information except what is necessary for the main task and preventing any secondary inference on the data, we introduce Siamese fine-tuning. We evaluate the privacy benefits of this approach based on the information exposed to the cloud service. We also assess the local inference cost of different layers on a modern handset. Our evaluations show that by using Siamese fine-tuning and at a small processing cost, we can greatly reduce the level of unnecessary, potentially sensitive information in the personal data, and thus achieving the desired trade-off between utility, privacy, and performance.
Tasks
Published	2017-03-08
URL	https://arxiv.org/abs/1703.02952v7
PDF	https://arxiv.org/pdf/1703.02952v7.pdf
PWC	https://paperswithcode.com/paper/a-hybrid-deep-learning-architecture-for-1
Repo
Framework

Jointly Learning Sentence Embeddings and Syntax with Unsupervised Tree-LSTMs


Title	Jointly Learning Sentence Embeddings and Syntax with Unsupervised Tree-LSTMs
Authors	Jean Maillard, Stephen Clark, Dani Yogatama
Abstract	We introduce a neural network that represents sentences by composing their words according to induced binary parse trees. We use Tree-LSTM as our composition function, applied along a tree structure found by a fully differentiable natural language chart parser. Our model simultaneously optimises both the composition function and the parser, thus eliminating the need for externally-provided parse trees which are normally required for Tree-LSTM. It can therefore be seen as a tree-based RNN that is unsupervised with respect to the parse trees. As it is fully differentiable, our model is easily trained with an off-the-shelf gradient descent method and backpropagation. We demonstrate that it achieves better performance compared to various supervised Tree-LSTM architectures on a textual entailment task and a reverse dictionary task.
Tasks	Natural Language Inference, Sentence Embeddings
Published	2017-05-25
URL	http://arxiv.org/abs/1705.09189v1
PDF	http://arxiv.org/pdf/1705.09189v1.pdf
PWC	https://paperswithcode.com/paper/jointly-learning-sentence-embeddings-and
Repo
Framework

Exact Learning of Lightweight Description Logic Ontologies


Title	Exact Learning of Lightweight Description Logic Ontologies
Authors	Boris Konev, Carsten Lutz, Ana Ozaki, Frank Wolter
Abstract	We study the problem of learning description logic (DL) ontologies in Angluin et al.‘s framework of exact learning via queries. We admit membership queries (“is a given subsumption entailed by the target ontology?") and equivalence queries (“is a given ontology equivalent to the target ontology?"). We present three main results: (1) ontologies formulated in (two relevant versions of) the description logic DL-Lite can be learned with polynomially many queries of polynomial size; (2) this is not the case for ontologies formulated in the description logic EL, even when only acyclic ontologies are admitted; and (3) ontologies formulated in a fragment of EL related to the web ontology language OWL 2 RL can be learned in polynomial time. We also show that neither membership nor equivalence queries alone are sufficient in cases (1) and (3).
Tasks
Published	2017-09-20
URL	http://arxiv.org/abs/1709.07314v1
PDF	http://arxiv.org/pdf/1709.07314v1.pdf
PWC	https://paperswithcode.com/paper/exact-learning-of-lightweight-description
Repo
Framework

A Study of All-Convolutional Encoders for Connectionist Temporal Classification


Title	A Study of All-Convolutional Encoders for Connectionist Temporal Classification
Authors	Kalpesh Krishna, Liang Lu, Kevin Gimpel, Karen Livescu
Abstract	Connectionist temporal classification (CTC) is a popular sequence prediction approach for automatic speech recognition that is typically used with models based on recurrent neural networks (RNNs). We explore whether deep convolutional neural networks (CNNs) can be used effectively instead of RNNs as the “encoder” in CTC. CNNs lack an explicit representation of the entire sequence, but have the advantage that they are much faster to train. We present an exploration of CNNs as encoders for CTC models, in the context of character-based (lexicon-free) automatic speech recognition. In particular, we explore a range of one-dimensional convolutional layers, which are particularly efficient. We compare the performance of our CNN-based models against typical RNNbased models in terms of training time, decoding time, model size and word error rate (WER) on the Switchboard Eval2000 corpus. We find that our CNN-based models are close in performance to LSTMs, while not matching them, and are much faster to train and decode.
Tasks	Speech Recognition
Published	2017-10-28
URL	http://arxiv.org/abs/1710.10398v2
PDF	http://arxiv.org/pdf/1710.10398v2.pdf
PWC	https://paperswithcode.com/paper/a-study-of-all-convolutional-encoders-for
Repo
Framework