Partners
Universita' Ca' Foscari di Venezia (Coordinator)
Link: www.unive.it
Contact person: Marcello Pelillo pelillo@dsi.unive.it
Description of the partner: Ca' Foscari University of Venice has an outstanding reputation for academic excellence in both teaching and research. Founded in 1868 as the first Italian school devoted to commerce and economics, it has grown and developed new relevant subject areas, including computer science. Within the Department of Computer Science, Prof. Pelillo leads since 1995 a research group in the areas of computer vision, pattern recognition, and neural computation, which presently includes two faculty members, a postdoctoral RA, two research students, and about ten undergraduate students. The group's research activity is focused primarily on graph-theoretic, optimization, and game-theoretic approaches, and in the interplay between continuous and combinatorial methods. The research spans a range of topics including grouping and segmentation, shape analysis and object recognition, structural matching and learning, and contextual pattern recognition. The group publishes regularly in top-level journals (e.g., PAMI, Neural Computation, IJCV, CVIU) and conferences (e.g., CVPR, ICCV, ECCV, NIPS), and has strong experience in leading and managing research projects, having coordinated in the past few years several national projects. Apart from the coordination of the whole project, the group's main contribution towards the consortium's effort will be on game-theoretic approaches and structural learning, whereby all group members have long experience both at the theoretical and the algorithmic/application levels. The group will also be heavily involved in the main application tasks of the project.
University of York
Link: www.york.ac.uk
Contact person: Edwin Robert Hancock erh@cs.york.ac.uk
Description of the partner: The Department of Computer Science at the University of York is one of the top computer science research departments in the UK, being one of just three (the others being Cambridge and Imperial College) to be awarded the top 5* classification in the 1996 and 2001 Research Assessment Exercises (RAE). Within the Department Professor Hancock and Dr Wilson lead a research group of some 20 faculty and researchers in the areas of computer vision and pattern recognition. The group has an excellent record of postgraduate training having graduated a total of some 30 completed PhD's over the past 15 years and has published large volumes of material in the top venues in the field (TPAMI, TIP, IJCV, CVIU, ICCV, ECCV, CVPR and NIPS). They have been principal or co-investigators for ten EPSRC funded research projects, the five most recently completed have received the top (outstanding/internationally leading) grade. The group focuses on problems of machine learning in computer vision. Current work is focused on learning structural descriptions of shape using information theoretic and graph-spectral techniques. There is also an extensive programme of work on geometric representation of surface shape, and the recovery of 3D shape from single 2D images. The group will be responsible for the work related to graph-embedding and spectral geometry, and it will be involved in the large-scale application tasks of the project.
Technische Universiteit Delft
Link: www.tudelft.nlk
Contact person: Robert P.W. Duin r.duin@ieee.orgk
Description of the partner: The TU Delft partner is the pattern recognition research team of the Information and Communication Theory section (10 permanent and 40 temporary researchers) of the Faculty of Electrical Engineering, Mathematics and Computer Science of Delft University of Technology (TU Delft) in the Netherlands. The pattern recognition research is headed by Prof. M. J. T. Reinders, who is also heading several bio-informatics projects. Dr. Robert Duin supervises the fundamental projects in pattern recognition. This involves studies on new representations for real-world objects, e.g. based on graphs, (dis)similarities and features, as well as new procedures for generalization, such as one-class classifiers and class-prior independent optimization strategies. The TU Delft team will be headed by Dr. Duin. It will be responsible for the project research with respect to the circumstances under which non-Euclidean distances arise. This team will investigate the best ways of using or correcting the data, depending on the circumstances. Dr. Duin supervised in total 15 PhD students, including two on the topic of learning from (dis)similarity data.
Instituto Superior Tecnico
Link: www.ist.utl.pt
Contact person: Mario Figueiredo mario.figueiredo@lx.it.pt
Description of the partner: Instituto Superior Tecnico (IST) is the engineering school of the Technical University of Lisbon; created in 1911, it is the largest (more than 10000 students and 700 faculty) and most reputed school of engineering, science, and technology in Portugal. The participation of IST in this project will be led by Prof. Mario Figueiredo, of the Communication Theory and Pattern Recognition (CTPR) group, of the Department of Electrical and Computer Engineering of IST. The CTPR group includes six faculty members and almost twenty graduate students, and carries out research activities in communication and information theory, signal processing, pattern recognition, machine learning, computer vision, image processing, medical imaging, remote sensing, and related areas. The group publishes regularly in top international journals (such as IEEE-TIP, IEEE-PAMI, IEEE-TGRS, IEEE-TAES, IEEE-JSAC, IEEE-TMI) and in top international conferences (such as IEEE-ICASSP, IEEE-ICIP, IEEE-PLANS, ION, NIPS, CVPR, ICPR, IEEE-VTC, IEEE-GLOBECOM, IEEE-WSSP). Members of the group have been associate editors of top journals (such as IEEE-TIP, IEEE-TPAMI, IEEE-TMC, IEEE-TSS, PRL, SP) and have been members of program committees of numerous prestigious international conferences, which confirms their international reputation. The main contribution of the CTPR group of IT to this research project will be on the development and study of information-theoretic (dis)similarity measures, with the goal of developing universal (dis)similarities for structural data, as well as the development of methods to learn (dis)similarity measures directly from unlabeled data, within an unsupervised learning scenario. The group will also be involved in the large-scale application tasks of the project.
Universita' degli Studi di Verona
Link: www.univr.it
Contact person: Vittorio Murrino vittorio.murino@univr.it
Description of the partner: University of Verona has been founded at the beginning of the 1960s, as faculty of Economics and Commerce. Thanks to the precious support and strict collaboration of the main public and private governmental institutional representatives and thanks to the support of its expert teachers, Verona University has grown over the years to have seven faculties among which the Mathematics, Physics and Natural Sciences, under which the Department of Computer Science is located. The Department of Computer Science gave rise in these last 10 years to several research activity strands, such as Programming Languages, Information Systems and Multimedia (this last one well-represented by the Vision, Image Processing, and Sound (VIPS) lab (http://vips.sci.univr.it) being each one of them profitable under the point of view of the publications produced and the laurea degree courses offered. In particular, the VIPS laboratory has expertise in video analysis and event recognition, generative modelling, biomedical data analysis. VIPS has contributed to the literature in these areas both with theoretical results and practical applications. VIPS was involved in many international projects and the VIPS crew has a notable experience in carrying out and managing research projects funded by the European Commission (VENICE and ARROV RTD projects) as well as by national institutions and industrial organizations. VIPS has started a fruitful research on MRI-Brain scans analysis in collaboration with the WHO Collaborating Centre for Research and Training in Mental Health and Service Evaluation within the context of the Verona-Udine Brain Imaging and Neuropsychology Program (VUBINP). The main contributions within the SIMBAD project are: 1) Studying and developing novel generative kernels, aimed at modelling structured data; 2) coordinating the activities related to the MRI scan analysis application.
Eidgenoessische Technische Hochschule Zuerich
Link: www.ethz.ch
Contact person: Joachim M. Buhmann jbuhmann@inf.ethz.ch
Description of the partner: The Swiss Federal Institute of Technology is one of the leading universities in Europe and the computer science department with its coverage of all major areas in computer science enjoys an internationally outstanding reputation. Joachim M. Buhmann joined ETH as professor for Information Science and Engineering to head the research group in Machine Learning with applications in computational biology, computer vision and acoustics. The group provides core competences in statistical modelling and algorithm design for machine learning. The group has two research associates, one postdoctoral researcher and ten PhD students. Current research projects range from image categorization, image and video segmentation, computational neuroanatomy and medical image analysis for cancer pathology to proteomics and metabolomics research and network inference in bioinformatics. Dr. Buhmann's interest in the foundations of machine learning and statistics also address topics like robust optimization, validation of learning and statistical vs. algorithmic complexity in inference. The group's main contribution towards the consortium's effort will be on embedding approaches of dissimilarity data, learning data similarity and validations, as well as providing the testbed of tissue microarray analysis for cancer diagnosis and survival analysis. Group members have a profound experience in statistical modeling and algorithms design as well as expertise in machine learning applications.