{"title":"Learning Causal Networks Topology From Streaming Graph Signals","authors":"M. Moscu, Roula Nassif, Fei Hua, C. Richard","doi":"10.23919/EUSIPCO.2019.8902826","DOIUrl":"https://doi.org/10.23919/EUSIPCO.2019.8902826","url":null,"abstract":"Modern data analysis and processing tasks usually involve large sets of data structured by a graph. Typical examples include brain activity supported by neurons, data shared by users of social media, and traffic on transportation or energy networks. There are often settings where the graph is not readily available, and has to be estimated from data. This paper focuses on estimating a network structure capturing the dependencies among streaming graph signals in the form of a possibly directed, weighted adjacency matrix. Several works proposed centralized offline solutions to address this problem, without paying much attention to the distributed nature of networks. We start from a centralized setting and show how, by introducing a simple yet powerful data model, we can infer a graph structure from streaming data with a distributed online learning algorithm. Our algorithm is tested experimentally to illustrate its usefulness, and successfully compared to a centralized offline solution of the literature.","PeriodicalId":319233,"journal":{"name":"2019 27th European Signal Processing Conference (EUSIPCO)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129472719","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Large-scale Canonical Polyadic Decomposition via Regular Tensor Sampling","authors":"Charilaos I. Kanatsoulis, N. Sidiropoulos","doi":"10.23919/EUSIPCO.2019.8902959","DOIUrl":"https://doi.org/10.23919/EUSIPCO.2019.8902959","url":null,"abstract":"Tensor decomposition models have proven to be effective analysis tools in various applications, including signal processing, machine learning, and communications, to name a few. Canonical polyadic decomposition (CPD) is a very popular model, which decomposes a higher order tensor signal into a sum of rank 1 terms. However, when the tensor size gets big, computing the CPD becomes a lot more challenging. Previous works proposed using random (generalized) tensor sampling or compression to alleviate this challenge. In this work, we propose using a regular tensor sampling framework instead. We show that by appropriately selecting the sampling mechanism, we can simultaneously control memory and computational complexity, while guaranteeing identifiability at the same time. Numerical experiments with synthetic and real data showcase the effectiveness of our approach.","PeriodicalId":319233,"journal":{"name":"2019 27th European Signal Processing Conference (EUSIPCO)","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128233316","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Monaural Speech Separation with Deep Learning Using Phase Modelling and Capsule Networks","authors":"Toby Staines, Tillman Weyde, Oleksandr Galkin","doi":"10.23919/EUSIPCO.2019.8902655","DOIUrl":"https://doi.org/10.23919/EUSIPCO.2019.8902655","url":null,"abstract":"The removal of background noise from speech audio is a problem with high practical relevance. A variety of deep learning approaches have been applied to it in recent years, most of which operate on a magnitude spectrogram representation of a noisy recording to estimate the isolated speaking voice. This work investigates ways to include phase information, which is commonly discarded, firstly within a convolutional neural network (CNN) architecture, and secondly by applying capsule networks, to our knowledge the first time capsules have been used in source separation. We present a Circular Loss function, which takes into account the periodic nature of phase. Our results show that the inclusion of phase information leads to an improvement in the quality of speech separation. We also find that in our experiments convolutional neural networks outperform capsule networks at speech separation.","PeriodicalId":319233,"journal":{"name":"2019 27th European Signal Processing Conference (EUSIPCO)","volume":"53 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128252176","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Region-Based Relaxations To Accelerate Greedy Approaches","authors":"Clément Dorffer, C. Herzet, Angélique Dremeau","doi":"10.23919/EUSIPCO.2019.8902669","DOIUrl":"https://doi.org/10.23919/EUSIPCO.2019.8902669","url":null,"abstract":"We propose a low-computational strategy for the efficient implementation of the “atom selection step” in sparse representation algorithms. The proposed procedure is based on simple tests enabling to identify subsets of atoms which cannot be selected. Our procedure applies on both discrete or continuous dictionaries. Experiments performed on the standard “Gaussian deconvolution” problem show the computational gain induced by the proposed approach.","PeriodicalId":319233,"journal":{"name":"2019 27th European Signal Processing Conference (EUSIPCO)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129662712","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Topology Inference and Signal Representation Using Dictionary Learning","authors":"Mahmoud Ramezani-Mayiami, K. Skretting","doi":"10.23919/EUSIPCO.2019.8902344","DOIUrl":"https://doi.org/10.23919/EUSIPCO.2019.8902344","url":null,"abstract":"This paper presents a Joint Graph Learning and Signal Representation algorithm, called JGLSR, for simultaneous topology learning and graph signal representation via a learned over-complete dictionary. The proposed algorithm alternates between three main steps: sparse coding, dictionary learning, and graph topology inference. We introduce the “transformed graph” which can be considered as a projected graph in the transform domain spanned by the dictionary atoms. Simulation results via synthetic and real data show that the proposed approach has a higher performance when compared to the well-known algorithms for joint undirected graph topology inference and signal representation, when there is no information about the transform domain. Five performance measures are used to compare JGLSR with two conventional algorithms and show its higher performance.","PeriodicalId":319233,"journal":{"name":"2019 27th European Signal Processing Conference (EUSIPCO)","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127284783","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Anxo Tato, C. Mosquera, P. Henarejos, A. Pérez-Neira
{"title":"Neural Network Aided Computation of Generalized Spatial Modulation Capacity","authors":"Anxo Tato, C. Mosquera, P. Henarejos, A. Pérez-Neira","doi":"10.23919/EUSIPCO.2019.8903104","DOIUrl":"https://doi.org/10.23919/EUSIPCO.2019.8903104","url":null,"abstract":"Generalized Spatial Modulation (GSM) is being considered for future high-capacity and energy efficient terrestrial networks. A variant such as Polarized Modulation (PMod) has also a role in Dual Polarization Mobile Satellite Systems. The implementation of adaptive GSM systems requires fast methods to evaluate the channel dependent GSM capacity, which amounts to solve multi-dimensional integrals without closed-form solutions. For this purpose, we propose the use of a Multilayer Feedforward Neural Network and an associated feature selection algorithm. The resulting method is highly accurate and with much lower complexity than alternative numerical methods.","PeriodicalId":319233,"journal":{"name":"2019 27th European Signal Processing Conference (EUSIPCO)","volume":"3 6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127287954","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"A Continuum Model for Route Optimization in Large-Scale Inhomogeneous Multi-Hop Wireless Networks","authors":"Dene A. Hedges, J. Coon, Gaojie Chen","doi":"10.23919/EUSIPCO.2019.8903086","DOIUrl":"https://doi.org/10.23919/EUSIPCO.2019.8903086","url":null,"abstract":"Multi-hop route optimization in large-scale inhomogeneous networks typically requires the use of constrained optimization tools, yielding processing complexity that scales like $O(N^{3}), N$ being the number of relays employed. Here, we propose an alternative approach to route optimization by considering the limit of infinite relay node density to develop a continuum model, which yields an optimized equivalent continuous relay path. The model is carefully constructed to maintain a constant connection density even though the node density scales without bound. This leads to a formulation for minimizing the end-to-end outage probability that can be solved using methods from the calculus of variations. We demonstrate the effectiveness of this new approach and its potential for reducing processing complexity by considering a network subjected to a point source of interference.","PeriodicalId":319233,"journal":{"name":"2019 27th European Signal Processing Conference (EUSIPCO)","volume":"63 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131028626","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Statistical Modeling of the Patches DC Component for Low-Frequency Noise Reduction","authors":"Antoine Houdard","doi":"10.23919/EUSIPCO.2019.8903131","DOIUrl":"https://doi.org/10.23919/EUSIPCO.2019.8903131","url":null,"abstract":"This work is devoted to patch-based image denoising. Assuming an additive white Gaussian noise (AWGN) on patches, we derive corresponding models on centered patches and on their DC components. Then we propose a strategy for improving a given path-based denoiser. Finally, we provide experiments with the recent denoising method HDMI that shows improvement of the denoising quality, particularly for residual low frequency noise.","PeriodicalId":319233,"journal":{"name":"2019 27th European Signal Processing Conference (EUSIPCO)","volume":"221 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132447068","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Visual Quality Analysis of Judder Effect on Head Mounted Displays","authors":"Saeed Mahmoudpour, P. Schelkens","doi":"10.23919/EUSIPCO.2019.8902665","DOIUrl":"https://doi.org/10.23919/EUSIPCO.2019.8902665","url":null,"abstract":"The extended field of view (FoV) offered by head mounted displays (HMD) increases the immersive experience, but it also introduces new visual quality challenges to be addressed. The judder artefact is a quality degradation factor that appears during object tracking and it is caused by eye movement relative to the display. As a first attempt to investigate the negative effect of judder in wide FoV applications, we built a new dataset of omnidirectional videos at different bitrates and judder severity levels. Two subjective tests were conducted to assess the quality in terms of perceived severity of judder in compressed video sequences. The outcomes provide new findings about the effect of judder on human perception. The results also give further understanding about the interaction between presentation quality and judder that can be utilized for developing objective models (c) to predict quality degradation in presence of judder.","PeriodicalId":319233,"journal":{"name":"2019 27th European Signal Processing Conference (EUSIPCO)","volume":"24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132486458","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Improved Lossless Image Compression Using Adaptive Image Rotation","authors":"P. Möller, T. Strutz","doi":"10.23919/EUSIPCO.2019.8902580","DOIUrl":"https://doi.org/10.23919/EUSIPCO.2019.8902580","url":null,"abstract":"State-of-the-art compression schemes generally have the ability to adapt themselves to the properties of the data to be compressed. This is a kind of learning process and requires the modification of internal variables influencing the treatment of subsequent data segments. In other words: the order of processing has an impact on the compression performance. In image compression, the order can be changed, for example, by rotating the input image by $90^{circ}, 180^{circ}$, or 270°. In application to lossless screen content compression, investigations with different compression schemes (LOCO-I, HEVC, FP8v3, and SCF) have shown that the rotation has a considerable impact on the compression performance. The difficulty, however, is to predict the best rotation. For the SCF (soft context formation) compression scheme, we have developed a method based on a tiny neural network that suggests a suitable rotation by evaluating basic colour properties of the image to be compressed. The compression can be improved by 0.7098% to 1.3817% depending on the image set tested.","PeriodicalId":319233,"journal":{"name":"2019 27th European Signal Processing Conference (EUSIPCO)","volume":"45 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132778843","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}