Keynote & Plenary Speakers for IWPR 2018
Prof. Junyu Dong, Ocean University of China, China
Prof. Junyu Dong received his BSc and MSc from the Department of Applied Mathematics at Ocean University of China in 1993 and 1999 respectively, and received his PhD in November 2003 in Heriot-Watt University, UK. He is currently a professor and the Deputy Dean of College of Information Science and Technology. His research interests include computer vision, underwater image processing and machine learning, with more than 10 research projects supported by NSFC, MOST and other funding agencies. He has published more than 100 journal and conference papers.
Title: Texture Generation with Perceptual Feature Learning
Abstract: This talk introduces a novel task of generating texture images based on perceptual feature learning. Although in the past procedural models and even commercial packages were commonly used for creating textures, texture generation from user-defined perceptual attributes or semantic descriptions has been rarely studied. Perceptual attributes, such as directionality, regularity, roughness together with semantic descriptions are important factors for human observers to describe a texture. Thus, the first question is how to learn these perceptual features; several methods including on both hand-crafted and deep learning features are introduced. Then we present a framework based on joint deep neural networks that combines adversarial training and perceptual feature regression for texture generation, while only random noise and user-defined perceptual attributes are required as input. In this framework, a preliminary trained convolutional neural network is essentially integrated with the adversarial framework, which can drive the generated textures to possess given perceptual attributes. An important aspect of the proposed model is that, if we change one of the input perceptual features, the corresponding appearance of the generated textures will also be changed. We designed several experiments to validate the effectiveness of the proposed method. The results show that the proposed method can produce high quality texture images with desired perceptual properties. Demonstrations based on mobile applications and virtual reality are also provided.
Prof. Min-Ling Zhang, Southeast University, China
Min-Ling Zhang received the BSc, MSc, and PhD degrees in computer science from the Department of Computer Science and Technology, Nanjing University, China, in 2001, 2004 and 2007, respectively. Currently, he is a Professor at the School of Computer Science and Engineering, Southeast University, China. His main research interests include machine learning and data mining. In recent years, Dr. Zhang has served as the General Co-Chairs of ACML'18, Program Co-Chairs of ACML'17, PRICAI'16, Area Chair/Senior PC member of IJCAI'18/'17, AAAI'18/'17, ICDM'18/'17, PAKDD'16/'15, PC member of KDD'18/'17, ICML'18/'17, NIPS'17/'16, etc. He is also on the editorial board of ACM Transactions on Intelligent Systems and Technology, Neural Networks, Frontiers of Computer Science, Science China Information Sciences, etc., and one of the guest editors for the Machine Learning Journal special issue on learning from multi-label data. Dr. Zhang is the secretary-general of the CAAI (Chinese Association of Artificial Intelligence) Machine Learning Society, standing committee member of the CCF (China Computer Federation) Artificial Intelligence & Pattern Recognition Society. He is also the recipient of the CCF-IEEE CS Young Scientist Award (2016).
张敏灵分别于2001年、2004年、2007年于南京大学计算机科学与技术系获学士、硕士和博士学位，现为东南大学计算机科学与工程学院教授。主要研究领域为机器学习、数据挖掘。近年来应邀担任第十届亚洲机器学习会议（ACML'18）大会主席，第九届亚洲机器学习会议（ACML'17）、第十四届亚太人工智能国际会议（PRICAI'16）程序主席，以及IJCAI'18/'17, AAAI'18/'17, ICDM'18/'17, PAKDD'16/'15等国际会议领域主席/高级程序委员、KDD'18/'17, ICML'18/'17, NIPS'17/'16等国际会议程序委员等。任《ACM Transactions on Intelligent Systems and Technology》、《Neural Networks》、《Frontiers of Computer Science》、《Science China Information Sciences》等期刊编委，以及《Machine Learning》等期刊客座编辑。现任中国计算机学会人工智能与模式识别专委会常务委员、中国人工智能学会机器学习专委会秘书长等。曾获CCF-IEEE CS青年科学家奖（2016年度）等。
Title: Binary relevance for multi-label learning
Abstract: In recent years, multi-label learning has become one of the important research topics in machine learning community. Binary relevance (BR) is arguably the most popular approach towards multi-label learning, which decomposes the multi-label learning problem into a number of independent binary classification problems, one per category. In view of the well-known weakness of BR, i.e. ignorance of label correlations, a number of enhanced versions of BR have been developed in recent years by endowing BR with the ability of label correlations exploitation. Nonetheless, in addition to label correlations exploitation, there are several factors which need to be considered to make BR-based approach work effectively. In this talk, I will introduce some of our recent progresses on BR-based multi-label learning.In recent years, multi-label learning has become one of the important research topics in machine learning community. Binary relevance (BR) is arguably the most popular approach towards multi-label learning, which decomposes the multi-label learning problem into a number of independent binary classification problems, one per category. In view of the well-known weakness of BR, i.e. ignorance of label correlations, a number of enhanced versions of BR have been developed in recent years by endowing BR with the ability of label correlations exploitation. Nonetheless, in addition to label correlations exploitation, there are several factors which need to be considered to make BR-based approach work effectively. In this talk, I will introduce some of our recent progresses on BR-based multi-label learning.
Prof. Dr. Xiaoyi Jiang, University of Munster, Germany
Journal of Pattern Recognition and Artificial Intelligence. In addition, he also serves on the Advisory Board and Editorial Board of several journals, including the IEEE Transactions on Medical Imaging, International Journal of Neural Systems, Pattern Analysis and Applications, and Pattern Recognition. Prof. Jiang is a Fellow of IAPR.
Title: Consensus learning
Abstract: Combining multiple models into one consensus model helps among others reduce the uncertainty in the initial models. Consensus learning can be formulated in arbitrary problem domains, either in an informal or a formal manner. In this talk the focus will be given to the formal framework of so-called generalized median computation. The concept of this framework and related computation algorithms will be presented. A variety of applications in pattern recognition will be shown to demonstrate the power of consensus learning.
Prof. Pedro Furtado, University of Coimbra, Portugal
Pedro Furtado is Professor at University of Coimbra UC, Portugal, where he teaches courses in both Computer and Biomedical Engineering. He has more than 25 years experience in both teaching, doing research and supervising industry projects. He has a broad interest in computer science subjects, with the main focus being on performance and scalability qualities of systems, and more recently also on image recognition for medical applications. Pedro did a lot of research in data warehousing and analytics, bigdata, data mining, and image recognition for medical applications. He also worked in assistive technologies, applying image recognition and mobile technologies to healthcare scenarios. Pedro has about 200 papers published in international conferences and journals, books published and several research collaborations with both industry and academia. In the last years, Pedro has spent time as visiting scholar in some of the most prestigious universities in the world, and collaborating with non-profit institutions. Besides a PhD in Computer Engineering from U. Coimbra (UC) (2000), Pedro Furtado also holds an MBA from Universidade Catolica Portuguesa (UCP) (2004).
Title: On BigData, Object Recognition and Medical ImagingAbstract: Object recognition performance near to the human visual systems (HSV) is a long lasting research goal. Is it bigdata, massive processing or deeper semantics that will win the challenge? Or maybe something in-between? In this keynote we review some works and results in recognition and classification of medical and food images. We also review bigdata capabilities and technologies, and some keypoints from deep learning. Then we go back to the grand challenge of object recognition and preview our own present and future work on the subject.
Prof. Xudong Jiang, Nanyang Technological University, Singapore
Prof. Xudong Jiang received the B.Sc. and M.Sc. degree from the University of Electronic Science and Technology of China, and received the Ph.D. degree from Helmut Schmidt University Hamburg, Germany. From 1986 to 1993, he worked as Lecturer at UESTC where he received two Science and Technology Awards from the Ministry for Electronic Industry of China. He was a recipient of the German Konrad-Adenauer Foundation young scientist scholarship. From 1993 to 1997, he was with Helmut Schmidt University Hamburg, Germany as scientific assistant. From 1998 to 2004, He worked with the Institute for Infocomm Research, A*Star, Singapore, as Senior Research Fellow, Lead Scientist and appointed as the Head of Biometrics Laboratory where he developed an software that achieved the fastest and the second most accurate fingerprint verification in the International Fingerprint Verification Competition (FVC2000). He joined Nanyang Technological University, Singapore as a faculty member in 2003 and served as the Director of the Centre for Information Security from 2005 to 2011. Currently, Dr Jiang is a tenured Association Professor in Nanyang Technological University. Dr Jiang has published over 120 research papers, including 20 papers in top IEEE journals: TPAMI, TIP, TSP and SPM, which are well-cited on Web of Science. He is also an inventor of 7 patents (3 US patents). Dr Jiang is a senior member of IEEE, elected voting member of IFS technical committee of IEEE Signal Processing Society, Associate editor of IEEE Signal Processing Letters and IET Biometrics. He has been serving as General Chair, Technical Program Committee Chair, Keynote Speaker and Session Chair of multiple international conferences. His research interest includes pattern recognition, computer vision, machine learning, image analysis, signal processing, machine learning and biometrics.
Title: Data-driving Pattern recognition: from Subspace Approach and Sparse Coding to Deep Learning
Prof. Julian FIERREZ, Universidad Autonoma de Madrid, Spain
Julian Fierrez received the M.Sc. and the Ph.D. degrees in telecommunications engineering from Universidad Politecnica de Madrid, Spain, in 2001 and 2006, respectively. Since 2002 he has been affiliated with the Biometric Recognition Group (ATVS), first at Universidad Politecnica de Madrid, and since 2004 at Universidad Autonoma de Madrid, where he is currently an Associate Professor. From 2007 to 2009 he was a visiting researcher at Michigan State University in USA under a Marie Curie fellowship. His research interests include general signal and image processing, pattern recognition, and biometrics, with emphasis on signature and fingerprint verification, multi-biometrics, biometric databases, system security, and forensic applications of biometrics. Dr. Fierrez is actively involved in multiple EU projects focused on biometrics (e.g. TABULA RASA and BEAT), has attracted notable impact for his research (more than 6,000 citations with h-index = 42 in Google Scholar), and is the recipient of a number of distinctions, including: Best Paper Awards at AVBPA 2003, ICB 2006, ICPR 2008, and ICB 2015; Best PhD Thesis Award in Computer Vision and Pattern Recognition in 2005-2007 by the IAPR Spanish liaison (AERFAI), EBF European Biometric Industry Award 2006, EURASIP Best PhD Award 2012, Medal in the Young Researcher Awards 2015 by the Spanish Royal Academy of Engineering, and the Miguel Catalan Award to the Best Researcher under 40 in the Community in Madrid in the general area of Science and Technology.
Title: Recent Advances in Face and Body BiometricsAbstract: Nowadays, biometric systems can achieve satisfactory performance in many applications, especially in controlled scenarios. When it comes to less controlled scenarios, these systems are affected by plenty of different variability sources that greatly hinder their performance. In the particular case of face recognition, these variability sources are related to changes in pose, illumination, expression, occlusions, distance to the camera, low resolution, and so forth. This talk is focused on the search of additional identity clues to enhance the performance of biometric systems. In the context of unconstrained scenarios where the main biometric trait is the face, researchers have shown recently the benefits in person recognition that can be obtained by combining biometrics with other kinds of soft biometric data, like gender, or some kind of silhouette information. Finally, motivated by the fact that faces and bodies are equally salient and available in unconstrained scenarios at a distance, we will discuss the combined used of face and body-based biometrics, in practical scenarios for person identification and gender recognition.
Prof. Kin Hong Wong, The Chinese University of Hong Kong, Hong Kong
Prof. Wong Kin Hong is an Associate Professor of the Department of Computer Science and Engineering of the Chinese University of Hong Kong. He received a PhD from the Department of Engineering of the University of Cambridge. His major research interest is in 3-D computer vision especially in pose estimation, structure from motion and tracking. He has investigated and developed many useful techniques in computer vision such as the four-point pose estimation algorithm and Kalman-trifocal pose estimation methods which are useful in many application areas such as automatic driving and virtual reality. He is also interested in pattern recognition, embedded applications, and computer music.
Title: Applying computer vision in wearable computer and virtual reality systemsAbstract: In this talk I will discuss various projects carried out in the Chinese University of Hong Kong (CUHK) on computer vision and machine learning. First, I will talk about research of developing an instant language assistant in a form of a wearable computer that can help users to translate foreign printed words. When the user is pointing to the word using his/her finger, the system will read out the word and translate its meaning instantly. I will also discuss a system that uses a robust ARUCO marker for applications in virtual reality. Since machine learning has shown to be very promising in many pattern recognition applications, we have conducted research using neural networks for edge detection, object tracking, and pose estimation. Theories together practical issues encountered in these research projects will be elaborated in the talk.
Prof. Aythami Morales, Universidad Autonoma de Madrid, Spain
Biography: Aythami Morales Moreno received his M.Sc. degree in Telecommunication Engineering in 2006 from Universidad de Las Palmas de Gran Canaria. He received his Ph.D degree from La Universidad de Las Palmas de Gran Canaria in 2011. He performs his research works in the BiDA Lab - Biometric and Data Pattern Analytics Laboratory at Universidad Autonoma de Madrid, where he is currently an Associate Professor. He has performed research stays at the Biometric Research Laboratory at Michigan State University, the Biometric Research Center at Hong Kong Polytechnic University, the Biometric System Laboratory at University of Bologna and Schepens Eye Research Institute (Harvard Medical School). His research interests include pattern recognition, computer vision, machine learning and biometrics signal processing. He is author of more than 70 scientific articles published in international journals and conferences. He has received awards from ULPGC, La Caja de Canarias, SPEGC, and COIT. He has participated in 7 National and European projects in collaboration with other universities and private entities such as UAM, UPM, EUPMt, Accenture, Unión Fenosa, Soluziona,...
Title: Behavioral Biometrics based on Human-Device InteractionAbstract: Behavioral biometrics emerge as a feasible way to model human interaction with electronic devices (smartphones, computers, tablets). Behavioral patterns can be used to statistical modelling users in terms of identity, age, and actions by means of existing sensors (mouse, keypad, touch screen…). Main advantages of behavioral biometrics are the transparent modelling and continuous authentication. Some of the main drawbacks are the high intra-class variability, short utterances and limited performance. This keynote will survey some of the most promising behavioral biometrics including its main challenges, applications and research opportunities.