The tutorial presents a rigorous formulation of the change and anomaly detection problems, which fits many signal/image analysis techniques and applications, including sequential monitoring and detection by classification. The tutorial describes in detail the most important approaches in the literature, following the machine-learning perspective of supervised, semi-supervised and unsupervised monitoring tasks. Particular emphasis will be given to: i) issues raising in multivariate settings, and ii) change/anomaly detection methods that use learned models, which are often adopted to handle signals and images, together with domain-adaptation techniques. The tutorial is accompanied by various examples where change/anomaly detection algorithms are applied to solve real world problems, including ECG monitoring in wearable devices, image analysis to detect defects in industrial manufacturing, and fraud detection in credit card transactions
Tutorial website: Giacomo Boracchi "Change and Anomaly Detection in Signals, Images, and General Data Streams"
Change and anomaly detection problems are ubiquitous in engineering. The prompt detection of changes and anomalies is often a primary concern, as they provide precious information for understanding the dynamics of a monitored process, and for activating suitable countermeasures. Changes, for instance, might indicate an unforeseen evolution of the process generating the data, or a fault in a machinery. Anomalies are typically considered the most informative samples, as for instance arrhythmias in an ECG tracing or frauds in a stream of credit card transactions. Not surprisingly, detection problems in time series/images/videos have been widely investigated in the signal processing community, in application scenarios that range from quality inspection to health monitoring.
The tutorial presents a rigorous formulation of the change and anomaly detection problems, which fits many signal/image analysis techniques and applications, including sequential monitoring and detection by classification. The tutorial describes in detail the most important approaches in the literature, following the machine-learning perspective of supervised, semi-supervised and unsupervised monitoring tasks. Particular emphasis will be given to: i) issues raising in multivariate settings, where the popular approach of monitoring the log-likelihood will be demonstrated to loose power when data-dimension increases, and ii) change/anomaly detection methods that use learned models, which are often adopted to handle signals and images. The tutorial also illustrates how advanced learned models, like convolutional sparse representation and structured dictionaries, as well as domain-adaptation techniques, can be used to enhance detection algorithms. Finally, best practices for designing suitable experimental testbed will be discussed.
The tutorial is accompanied by various examples where change/anomaly detection algorithms are applied to solve real world problems. These include ECG monitoring in wearable devices, image analysis to detect defects in industrial manufacturing, and fraud detection in credit card transactions.
Giacomo Boracchi received the M.S. degree in Mathematics from the Università Statale degli Studi di Milano, Italy, and the Ph.D. degree in Information Technology at Politecnico di Milano, Italy, in 2004 and 2008, respectively. He was researcher at Tampere International Center for Signal Processing, Finland, during 2004-2005. Currently, he is an assistant professor at Dipartimento di Elettronica e Informazione, Politecnico di Milano.
His research interests encompass two different areas: computational intelligence and image analysis and enhancement. In particular, my research activity covers the following lines: learning methods for nonstationary environments, change/anomaly detection, computational imaging, and image restoration. In 2015 he received the IBM Faculty Award, and in 2016 the IEEE Trans. on Neural Networks and Learning Systems Outstanding Paper Award.
Dr. Azeddine BEGHDADI is Full Professor at the University of Paris 13 (Institut Galilée) Sorbonne Paris Cite since 2000 the director of L2TI laboratory (from 2010 to 2016). He received Maitrise in Physics and Diplôme d’Etudes Approfondies in Optics and Signal Processing from University Orsay-Paris XI (Equivalent: Masters of Sciences) in June 1982 and June 1983 respectively and the PhD in Physics (Specialism: Optics and Signal Processing) from University Paris 6 in June 1986. Dr. Beghdadi worked at different places during his PhD thesis, including "Laboratoire d’Optique des Solides" (University Pierre et Marie Curie - Paris 6) and "Groupe d’Analyse d’Images Biomédicales" (CNAM Paris). From 1987 to 1989, he has been appointed "Assistant Associé" (Lecturer) at University Paris 13. During the period 1987-1998, he was with LPMTM CNRS Laboratory working on Scanning Electron Microscope (SEM) materials image analysis. He published over than 260 international refereed scientific papers. He is a founding member of the L2TI laboratory. His research interests include image quality enhancement and assessment, image and video compression, bio-inspired models for image analysis and processing, and physics-based image analysis. Dr. Beghdadi is the founder and Co-Chair of the European Workshop on Visual Information Processing (EUVIP). Dr Beghdadi is a member of the editorial board of "Signal Processing: Image Communication" journal, Elsevier, EURASIP Journal on Image and Video Processing, the Journal of Electronic Imaging, and Mathematical Problems in Sciences Journal. Dr Beghdadi is a Senior member of IEEE, EURASIP member and member of the IEEE-MMTC.
Over the last fifty years, research and development in image/video processing have been driving advances in many high-tech areas, including medical and scientific imaging, digital cinema, computational photography, biometrics, remote sensing, among others. With the development of new imaging modalities and multimedia products, many new approaches have been proposed in this field of research. The research focus is however steadily shifting towards developing new mathematical models rather than understanding the image signal as a physical quantity and its interaction with the observer; unfortunately, the human user is often ignored in the image processing chain. Whereas, in many applications, such as diagnosis, recognition and evaluation, the human observer plays a prominent role in decision-making, based on visual assessment of images. Therefore, exploiting knowledge about the Human Visual System (HVS) in the design of multimedia processing techniques appears as a promising direction. Indeed, in many applications, it has been shown that by employing multi-channel models and by taking into account properties and limitations of the human visual system, visual multimedia content can be processed and analyzed more efficiently. A lot of current research based on the use of HVS models has been developed. This tutorial provides an overview of the most recent trends and the future research in image and video processing in a common perceptually based computational framework. The most relevant characteristics and properties of the HVS are then presented on the light of the recent findings in a concise and practical way.
This tutorial will be organized in three parts:
Abstract. In this tutorial the mathematical framework of light field signal processing is presented. Then, processing techniques for image refocusing and depth estimation are reviewed. The tutorial is organized as follows. In the first part the image formation in plenoptic cameras is presented. In these cameras multiple views of a scene are captured in a single shot by means of a micro-lens array placed on the focal point of the first camera lens, in front of the imaging sensor. In the second part, algorithms for image refocusing on both spatial and frequency domain are presented and their performance illustrated by means of numerical examples. Finally, in the third part, methods for the estimation of the depth field are introduced with focus on The local Maximum Likelihood estimation of the depth field based on Epipolar Plane Images.
Biography. Alessandro Neri is full professor in Telecommunications at the Engineering Department of the University "Roma Tre" of Rome, Italy. In 1977 he received the Doctoral Degree cum laude in Electronic Engineering from the University of Rome "La Sapienza". In 1987 he joined the INFOCOM Department of the University of Rome "La Sapienza" as Associate Professor in Signal and Information Theory at the Engineering Faculty. In November 1992 he joined the Electronic Engineering Department of the University of Roma TRE as Associate Professor in Electrical Communications, and became full professor in Telecommunications in semptember 2001. He is currently teaching Digital Communications, Information Theory, and Navigation and Localization Systems, at the Engineering Department of Roma Tre. His research activity has mainly been focused on Information Theory, Detection and Estimation Theory, Digital Signal Processing, and Image Processing and their applications to both telecommunications systems, navigation, and remote sensing. He is author of more than 300 publications.