|
|
 |
|
IEEE
Computational Intelligence Society
Ottawa
Chapter
|
|
|
|
Recent Meetings
2009
|
Q. J. Zhang |
|
2008
|
James M. Keller |
|
Marcel Turcotte |
|
Moufid Harb |
|
2007
|
Oana Frunza |
|
Jacek M. Zurada |
|
Stan Matwin |
|
Monique Frize |
|
2006
|
Evangelia Micheli-Tzanakou |
|
Diana Inkpen |
|
Jian Pei |
|
Jean-Philippe Thivierge |
|
2005
|
Mark Fiala |
|
Adrian D.C. Chan
|
|
Wail Gueaieb
|
|
B. John Oommen
|
|
2004
|
Dmitry Gorodnichy
|
|
Julio J. Valdés
|
|
Ana-Maria Cretu
|
|
Peter Turney
|
|
2003
|
Emil M. Petriu
|
|
Nicolas D. Georganas
|
|
Date |
Tuesday February 10, 2009 |
Time |
12:00-13:30 |
Location |
Room 5084, SITE Building, University of Ottawa |
Title |
Neural Networks for High-Frequency Electronic Modeling and Design |
Speaker |
Q. J. Zhang, Professor (IEEE Fellow) |
| |
Carleton University |
| |
Department of Electronics |
| |
|
| |
|
Abstract
Recent advances in the application of Artificial Neural Networks (ANN) to radio-frequency (RF) and microwave design created an exciting direction of computer-aided modeling and design of high-frequency electronics. ANNs are trained to learn the high-frequency behavior of electronic components, and trained ANNs can be used as models for high-level electronic design. The ANN models are much faster than detailed electromagnetic/physics based models of electronic components, and more accurate than conventional empirical/equivalent circuit models. It leads to substantial increase in modeling accuracy, speed, and flexibility. Applications are being made in modeling and design of passive and active RF/microwave electronic components and circuits, high-speed VLSI interconnects, printed antennas, LTCC circuits, semiconductor devices, measurement standards, filters, amplifiers, mixers and so on. Automated model generation algorithms integrating data generation and ANN training are being developed. Knowledge based neural networks exploiting prior knowledge such as empirical/semi-analytical models are being introduced in microwave computer-aided design (CAD). This leads to new level of CAD methodologies combining equivalent circuit/empirical models, electromagnetic/physics simulation and behavioral modeling with ANN and optimization algorithms for fast and accurate design of high-frequency circuits and systems. This talk presents a review of the state of the art in these emerging directions. The presentations highlight implementable methodologies for automated modeling and design of high-frequency electronic components, circuits and systems. The presentation covers fundamental concepts and methodologies, industrial applications, and future trends in R&D. |
| |
Speaker Bio
Q.J. Zhang received the B.Eng. degree from the Nanjing University of Science and Technology, Nanjing, China in 1982, and the Ph.D. Degree in Electrical Engineering from McMaster University, Hamilton, Canada, in 1987. He joined the Department of Electronics, Carleton University, Ottawa, Canada in 1990 where he is presently a Professor.
His research interests are modeling, optimization and neural networks for high-speed/high-frequency electronic design, and has over 200 publications in the area. He is an author of the book Neural Networks for RF and Microwave Design (Boston: Artech House, 2000), and a coeditor of Modeling and Simulation of High-Speed VLSI Interconnects (Boston: Kluwer, 1994). He is a contributor to Encyclopedia of RF and Microwave Engineering, (New York: Wiley, 2005), Fundamentals of Nonlinear Behavioral Modeling for RF and Microwave Designs, (Boston: Artech House, 2005), Tutorials on Emerging Methodologies and Applications in Operations Research, (New York: Springer, 2005), and Analog Methods for Computer-Aided Circuit Analysis and Diagnosis, (New York: Marcel Dekker, 1988). He was a Guest co-Editor for the Special Issue on High-Speed VLSI Interconnects for the International Journal of Analog Integrated Circuits and Signal Processing (Boston: Kluwer, 1994), and twice a Guest Editor for the Special Issues on Applications of ANN to RF and Microwave Design for the International Journal of RF and Microwave CAE (New York: Wiley, 1999, 2002).
Dr. Zhang is on the editorial board of the IEEE Transactions on Microwave Theory and Techniques, the International Journal of RF and Microwave CAE, and the International Journal of Numerical Modeling. He is an Associate Editor for the Journal of Circuits, Systems and Computers. He is a member of the Technical Committee on CAD of the IEEE MTT Society. He is a Fellow of the IEEE, and a Fellow of the Electromagnetics Academy..
|
Date |
Thursday December 18, 2008 |
Time |
10:30-12:00 |
Location |
Building M-50, NRC Auditorium, 1200 Montreal Road |
Title |
Soft Computing for Sensor and Algorithm Fusion |
Speaker |
James M. Keller , Professor |
| |
University of Missouri-Columbia |
| |
Electrical and Computer Engineering Department |
| |
|
| |
|
The talk will be followed by the annual meeting of the chapter |
Abstract
Sensor and algorithm fusion is playing an increasing role in many application domains. As detection and recognition problems become more complex and costly (for example, landmine detection and automatic activity monitoring), it is apparent that no single source of information can provide the ultimate solution. However, complementary information can be derived from multiple sources. Given a set of outputs from constituent sources, there are many frameworks within which to combine the pieces into a more definitive answer. This tutorial will focus on the fusion of multiple partial confidence values within the framework of fuzzy set theory.
So, the question then becomes: what methodology do we use to combine partial decision information? There are many choices, but I will focus on the use of fuzzy set theoretic mechanisms to fuse confidence from multiple sources. Two general approaches will be considered, fuzzy integrals and fuzzy logic rule-based systems. Fuzzy integrals have a long history and have been studied in the context of pattern recognition and information fusion for several years being first introduced for this purpose by Tahani and Keller in 1990. Fuzzy integrals combine the objective evidence supplied by each information source with the expected worth of each subset of information sources (via a fuzzy measure) to assign confidence to hypotheses or to rank alternatives in decision making. This is a nonlinear combination of information and the worth of the information for the decision in question, dealing with the uncertainty in both forms of data. Different fuzzy measures yield different integration operations, including averaging, linear combinations of order statistics, and many others. Measures can be found by heuristic assignment or via training algorithms. New results with discriminative training will be discussed. Next, a fusion system based on a linguistic extension of the Choquet fuzzy integral will be shown. The uncertainty in the data is now expressed as a linguistic vector, i.e., a vector of fuzzy sets. The linguistic Choquet integral is used to fuse both position and confidence uncertainty in the landmine detection scenario.
Fuzzy logic rule-based systems provide another mechanism to fuse together the results of different features, classification algorithms and sensors. Such a system employs rules much like those that a human expert might derive. Again, uncertainty in the component parts is modeled by linguistic variables taking on fuzzy sets as values. I will describe the application of fuzzy rule-based classifiers in image processing and landmine detection. |
| |
Speaker Bio
James M. Keller received the Ph.D. in Mathematics in 1978. He holds the University of Missouri Curators’ Professorship in the Electrical and Computer Engineering and Computer Science Departments on the Columbia campus. He is also the R. L. Tatum Professor in the College of Engineering. His research interests center on computational intelligence: fuzzy set theory and fuzzy logic, neural networks, and evolutionary computation with a focus on problems in computer vision, pattern recognition, and information fusion including bioinformatics, spatial reasoning in robotics, geospatial intelligence, sensor and information analysis in technology for eldercare, and landmine detection. His industrial and government funding sources include the Electronics and Space Corporation, Union Electric, Geo-Centers, National Science Foundation, the Administration on Aging, The National Institutes of Health, NASA/JSC, the Air Force Office of Scientific Research, the Army Research Office, the Office of Naval Research, the National Geospatial Intelligence Agency, and the Army Night Vision and Electronic Sensors Directorate. Professor Keller has coauthored over 300 technical publications.
Jim is a Fellow of the Institute of Electrical and Electronics Engineers (IEEE) for whom he has presented live and video tutorials on fuzzy logic in computer vision, is an International Fuzzy Systems Association (IFSA) Fellow, is a national lecturer for the Association for Computing Machinery (ACM), is an IEEE Computational Intelligence Society Distinguished Lecturer, and is a past President of the North American Fuzzy Information Processing Society (NAFIPS). He received the 2007 Fuzzy Systems Pioneer Award from the IEEE Computational Intelligence Society. He finished a full six year term as Editor-in-Chief of the IEEE Transactions on Fuzzy Systems, is an Associate Editor of the International Journal of Approximate Reasoning, and is on the editorial board of Pattern Analysis and Applications, Fuzzy Sets and Systems,International Journal of Fuzzy Systems, and the Journal of Intelligent and Fuzzy Systems. He is currently the Vice President for Publications of the IEEE Computational Intelligence Society. He was the conference chair of the 1991 NAFIPS Workshop, program co-chair of the 1996 NAFIPS meeting, program co-chair of the 1997 IEEE International Conference on Neural Networks, and the program chair of the 1998 IEEE International Conference on Fuzzy Systems. He was the general chair for the 2003 IEEE International Conference on Fuzzy Systems.
|
Date |
Thursday July 3, 2008 |
Time |
11:00-12:00 |
Location |
Room 379, Building M-50, NRC Auditorium, 1200 Montreal Road |
Title |
Applying Relational Learning to Structural Molecular Biology Problems |
Speaker |
Marcel Turcotte, Professor |
| |
University of Ottawa |
| |
School of Information, Technology and Engineering |
| |
|
| |
|
Abstract
I will present our work on applying relational learning to discover rules characterizing protein folds. Inductive Logic Programming (ILP) has been chosen for its ability to 1) discover relations, 2) represent background information, as well as 3) its expressiveness. Several representations for the background set have been explored, and the results have been interpreted in their biological context.
The rules that were automatically found are often similar to the descriptions found in SCOP (a database of protein folds) or the published scientific literature. Finally, I will conclude presenting some future applications, in particular, for the determination of protein function from structure information.
This is a joint work with M.J.E. Sternberg and S. Muggleton from Imperial College London/UK. |
| |
Speaker Bio
I (Marcel Turcotte) completed a Ph.D. at the University of Montreal, Canada, under the supervision of Guy Lapalme and Robert Cedergren. I was then a postoctoral fellow at the University of Florida, USA, where I worked with Steven Benner on evolutionary-based approaches to protein secondary structure prediction. I then moved to the United Kingdom to work in the Biomolecular Modelling Laboratory (Mike Sternberg, head) of the Imperial Cancer Research Fund. In 2000, I returned to Canada where I joined the School of Information Technology and Engineering at the University of Ottawa.
|
Date |
Tuesday March 18, 2008 |
Time |
10:00-11:00 |
Location |
Room 5084, SITE Building, University of Ottawa |
Title |
Neural Networks for Environment Recognition and Local Navigation of a Mobile Robot |
Speaker |
Moufid Harb, Ph.D., Research Scientist |
| |
Larus Technologies Corp. |
| |
|
| |
University of Ottawa |
| |
School of Information, Technology and Engineering |
| |
|
| |
|
Abstract
This presentation will focus on a computer based design and test of three neural controllers for local navigation, and another two neural networks for environmental recognition, fed off-line by a simulated model of a laser range-finder. These neural networks are the major components of a control system that performs a global neural navigation of a mobile robot, which could be used to perform industrial missions within industrial environments. This control system can guide a mobile robot to track its predefined path to arrive to its final goal through a set of sub-goals, or autonomously plan its path to arrive to the desired final goal, and to avoid obstacles that are found along the way. The presentation will include simulation results and live demonstrations. |
| |
Speaker Bio
Moufid Harb (M'07) is a Research Collaborator with the School of Information Technology and Engineering at the University of Ottawa, and a Research Scientist with Larus Technologies. He received his Bachelor of Eng. in Electrical Engineering in 1983 from Damascus University, his M.Sc. in Instrument Design and Application in 1994 from Manchester University, Institute of Science and Technology (UMIST), England-UK and his Ph.D. in Electrical Engineering in 2001 from Damascus University/Syria in collaboration with Ruher University of Bochum/Germany. His research interests include autonomous robotic navigation, sensor modeling and simulation, and intelligent systems. Dr. Harb is a member of the IEEE Ottawa Section. He is a member of the IEEE Instrumentation and Measurement Society, and the IEEE Computational Intelligence Society. He is currently a vice-chair of the IEEE Computational Intelligence Society - Ottawa Chapter.
|
Date |
Wednesday December 19, 2007 |
Time |
10:30-11:30 |
Location |
Room 5084, SITE Building, University of Ottawa |
Title |
Does pain hurt in both French and English? |
Speaker |
Oana Frunza, Ph.D. Candidate |
| |
University of Ottawa |
| |
Text Analysis and MAchine LEarning (TAMALE) Group |
| |
|
| |
|
The talk will be followed by the annual meeting of the chapter
|
Abstract
Cognates are pair of words in different languages similar in spelling and meaning. They can help a second-language learner on the tasks of vocabulary expansion and reading comprehension. False friends are pairs of words that have similar spelling but different meanings. Partial cognates are pairs of words in two languages that have the same meaning in some, but not all contexts. Detecting the actual meaning of a partial cognate in context can be useful for Machine Translation tools and for Computer-Assisted Language Learning tools.
In this talk I will present research that I have done for cognates and false-friends identification and partial cognate disambiguation tasks. I will describe the method that we propose to automatically classify a pair of words as cognates or false friends, and also a supervised and a semi-supervised method to disambiguate partial cognates between two languages. We applied all our methods to French and English, but they can be applied to other pairs of languages as well.
I will also present a tool that I built to annotate French texts with equivalent English cognates or false friends, in order to help a second-language learner. |
| |
Speaker Bio
Oana Frunza is currently a Ph.D. Candidate at University of Ottawa, Canada. She is doing research in Natural Language Processing and Machine Learning with Dr. Diana Inkpen. She has a Computer Science background form “Babes-Bolyai” University, Romania and a M.S.C. Diploma from University of Ottawa, Canada in Natural Language Processing and Machine Learning. Her main research is focused on automatic text classification, semantic representation and machine learning techniques applied to various text processing tasks. She is an author or co-author of papers that were published in prestigious international conferences.
|
Date |
Friday November 09, 2007 |
Time |
10:30-12:00 |
Location |
Building M-50, NRC Auditorium, 1200 Montreal Road |
Title |
Data Mining, Neural Networks and Rule Extraction |
Speaker |
Jacek M. Zurada, Distinguished University Scholar (IEEE Fellow) |
| |
University of Louisville |
| |
Computational Intelligence Laboratory |
| |
|
| |
|
Abstract
This lecture discusses paradigms of neurocomputing in context of effective data mining tasks such as data-driven modeling, feature extraction, dimensionality reduction, visualization, knowledge extraction and logic rule discovery. These tasks often involve handling of heterogenous, subjective, imprecise and noisy data. Of special importance here is the concept of dimensionality reduction of input data vectors. An approach is presented that leads to reduced models achieved through evaluation of sensitivity matrices of perceptron networks. When developing reduced models it is also useful to eliminate underutilized internal weights and, possibly, also neurons via pruning techniques. The concluding part of the talk reviews the potential of perceptron networks for producing understandable IF-THEN rules. |
| |
Speaker Bio
Dr. Jacek M. Zurada serves as the Distinguished University Scholar and Professor of Electrical and Computer Engineering at the University of Louisville, Louisville, Kentucky, USA. He is the author of several books such Introduction to Artificial Neural Systems, and co-editor of Computational Intelligence: Imitating Life, and of Knowledge Based Neurocomputing. He is also the author or co-author of more than 300 journal and conference papers in the area of neural networks and computational intelligence. In 1998-2003, Dr. Zurada was the Editor-in-Chief of IEEE Transactions on Neural Networks. In 2004-05 he served as the President of IEEE Computational Intelligence Society. He is an IEEE Fellow.
|
|
Date |
Tuesday July 17, 2007 |
Time |
13:30-14:30 |
Location |
Room 379, Building M-50, NRC Auditorium, 1200 Montreal Road |
Title |
|
Speaker |
Stan Matwin, Professor |
| |
University of Ottawa |
| |
School of Information, Technology and Engineering |
| |
|
| |
|
Abstract
In this talk we will discuss two current applications of Machine Learning, being developed at the Text Analysis and Machine Learning (TAMALE) research group at the University of Ottawa. The first application (joint work with Dr. D. Inkpen), in cooperation with TrialStat, targets screening of medical abstracts for a Systematic Review System. Systematic reviews are the basic tool of Evidence-based-medicine. We will describe the task, and outline the requirements and challenges a Machine Learning solution must meet. The second application is in the area of Digital games Based Learning. In a joint effort with Distil Interactive, we are using Machine Learning in acquiring profiles of different classes of people using digital games for skill certification. Here as well we will outline some of the requirements and challenges the task presents from the perspective of Machine Learning. Interestingly, some of the challenges are shared by both tasks and are among the challenges before the entire field. |
| |
Speaker Bio
Stan Matwin is a professor at the School of Information Technology and Engineering, University of Ottawa, where he directs the Text Analysis and Machine Learning (TAMALE) lab. His research is in machine learning, data mining, and their applications, as well as in technological aspects of Electronic Commerce. Author and co-author of 150 research papers, he has worked at universities in Canada, the U.S., Europe and Latin America, where in 1997 he held the UNESCO Distinguished Chair in Science and Sustainable Development. Former president of the Canadian Society for the Computational Studies of Intelligence (CSCSI) and of the IFIP Working Group 12.2 (Machine Learning). Founding Director of the Graduate Certificate in Electronic Commerce at University of Ottawa. Founding Director of the Information Technology Cluster of the Ontario Research Centre for Electronic Commerce. Chair of the NSERC Grant Selection Committee for Computer Science and member of the Board of Directors of Communications and Information Technology Ontario (CITO). Recipient of a CITO Champion of Innovation Award. Programme Committee Chair and Area Chair for a number of international conferences in AI and Machine Learning. Member of the Editorial Boards of the Machine Learning Journal, Computational Intelligence Journal, and the Intelligent Data Analysis Journal.
|
|
|
|
Date |
Monday February 07, 2007 |
Time |
17:30-19:00 |
Location |
Room 5084, SITE Building, University of Ottawa |
Title |
Information Technologies Applied to Medicine |
Speaker |
Monique Frize, Professor |
| |
Carleton University |
| |
Department of Systems and Computer Engineering |
| |
|
| |
University of Ottawa |
| |
School of Information, Technology and Engineering |
| |
|
| |
|
Abstract
The talk will present current research and development in biomedical engineering. Using machine learning and data mining techniques such as artificial neural networks and case-based reasoning, we design and attempt to improve the performance of these tools to predict premature births (before the 23rd week of gestation); we also predict complications for infants in intensive care; we are developing an automated technique to assess pain levels in babies and adults using an infra-red camera and imaging techniques. |
| |
Speaker Bio
Dr. Frize joins Carleton University, as a Professor in the Department of Systems and Computer Engineering, and the University of Ottawa, as a Professor in the School of Information Technology and Engineering, in July 1997.
Dr. Frize graduated with a Bachelor of Applied Science (Electrical Engineering), received an Athlone Fellowship and completed a Master's in Philosophy in Electrical Engineering (Engineering in Medicine) at Imperial College of Science and Technology in London (UK), a Master's of Business Administration at the Université de Moncton (New Brunswick), and a doctorate from Erasmus Universiteit in Rotterdam, The Netherlands.
Monique Frize worked as a clinical engineer for 18 years, initially at Hopital Notre-Dame in Montreal (1971-79), and then was appointed as Director of the Regional Clinical Engineering Service in Moncton, New Brunswick, providing services for seven hospitals in the South-Eastern region. Dr. Frize was also Research Associate in the Faculty of Science and Engineering at UniversitJ de Moncton and was the first Chair of the Division of Clinical Engineering for the International Federation of Medical and Biological Engineering (IFMBE). In December, l989, she was appointed the first holder of the Nortel-NSERC Women in Engineering Chair at the University of New Brunswick (Fredericton) and Professor in the Electrical Engineering department.
In 1992, Monique Frize received an Honorary Doctorate from the University of Ottawa (DU); in June 1993, a Ryerson Fellowship; in 1994, an Honourary Doctorate in Science (DSc) at York University; in 1995, an Honourary Doctorate in Engineering at Lakehead (DEng). She was inducted as a Fellow of the Canadian Academy of Engineering in 1992 and as Officer of the Order of Canada in October 1993. In 1995, Dr. Frize received the Second Historical Professional Achievement Award (jointly with Dr. Michael Shaffer) from the American College of Clinical Engineers, for her paper: "Clinical Engineering in today's hospital: Perspectives of the Administrator and the Clinical Engineer". In September 1996, Dr. Frize received the 6th Annual Meritas-Tabaret Award for career achievement from the Alumni Association of the University of Ottawa and the Advocacy Award presented by WITT (Women in Trades and Technology) in May 1997. Born in Montreal, Canada, Dr. Frize's mother tongue is French, and she is fluently bilingual. She is married to Peter Frize and they have a son, Patrick Nicholas. |
Date |
Monday December 04, 2006 |
Time |
10:30-12:00 |
Location |
Building M-50, NRC Auditorium, 1200 Montreal Road |
Title |
Lying, Deception and Face Familiarity with Visual Evoked Potentials |
Speaker |
Evangelia Micheli-Tzanakou, Professor |
| |
Rutgers University |
| |
Department of Biomedical Engineering |
| |
Director of Computational Intelligence Laboratories |
| |
|
| |
|
Abstract
The recent urgency for counter-terrorism and the fight to protect ones’ homeland are of grave concern to nations throughout the world. Finding an efficient process that could have screened passengers prior to boarding a flight or even a train may have derailed many devastating events. Scientific research has continuously proved that there is an explicit marker of neuronal activity that correlates with awareness, past experience, and short-term interactions from the well-known P300 peak of an evoked potential. This study examines the effects of memory recognition to certain key stimuli mixed with irrelevant variables in an effort to identify if a trained terrorist, per se, could not only be identified from a group of subjects, but also validate that the willingness to withhold information is out of the control of the individual; it is simple, one has no control in concealing their brain’s activity. Face familiarity is also examined through a series of experiments. |
| |
Speaker Bio
Dr. Tzanakou is Professor, Director of Computational Intelligence Laboratories (2000-present); Chair Biomedical Eng., Rutgers University (1990-00). Has published 250+ scientific papers. Authored/co-authored 4 books/edited proceedings, graduated 37 M.S. and Ph.D.'s; Founding Fellow, American Institute for Medical & Biological Eng., 1993; Member of Sigma Xi and Eta Kappa Nu; Honorary Member: British Brain Research Association; European Brain/Behavior Society. Awards: NJ Women of Achievement, 1995; Featured "Notable Twentieth Century Scientists," 1994; Achievement Award, Society of Women Engineers, 1992; Outstanding IEEE Branch Advisor/Counselor, 1985; Pioneer (IEEE Web page). Book Series Editor: Biomedical Engineering, Plenum/Klewer (1999-present); Biomedical Systems, IES Book series, CRC Press (1997-present); Editorial Board: IEEE Transactions on Nano-BioScience (2002-present); Biomedical Engineering On Line, (2001-present), IEEE Transactions on Biomedical Information Technology (1997-01); Intern. J. Adv. Computational Intelligence (1997-99); Advanced Computational Intelligence and Intelligent Informatics, (2000-present); International J. on Artificial Intelligence Tools, (2004-present). Associate Editor, IEEE Transactions on Neural Networks (2000-present), (1989-92). |
Date |
Friday November 17, 2006 |
Time |
17:30-19:00 |
Location |
Room 5084, SITE Building, University of Ottawa |
Title |
|
Speaker |
Diana Inkpen, Assistant Professor |
| |
University of Ottawa |
| |
School of Information Technology and Engineering |
| |
|
| |
|
Abstract
Browsing through large volumes of spoken audio is known to be a challenging task for end users. To alleviate this problem we can allow users to gist a spoken audio document by glancing over a transcript generated through Automatic Speech Recognition, or to implement information retrieval systems over the text transcribed by the speech recognizer.
Unfortunately, such transcripts typically contain many recognition errors which are highly distracting and make gisting more difficult. I present an approach that detects recognition errors by identifying words which are semantic outliers with respect to other words in the transcript. I investigate a wide range of evaluation measures and show that we can significantly reduce the number of errors in content words, with the trade-off of losing some good content words.
Also described are information retrieval experiments from a collection of spontaneous speech. I show comparative results for indexing the automatic transcripts as opposed to indexing the manual summaries and keywords available in the collection. |
| |
Speaker Bio
Dr. Diana Inkpen is a professor at the School of Information Technology and Engineering, University of Ottawa since July 2003. She obtained her doctorate in 2003 from University of Toronto, Department of Computer Science. She has a Masters in Computer Science and Engineering from the Technical University of Cluj-Napoca, Romania. Her research projects and publications are in the areas of Computational Linguistics and Artificial Intelligence, more specifically: Information Retrieval, Information Extraction, Natural Language Understanding, Natural Language Generation, Speech Processing, and Intelligent Agents for the Semantic Web.
Dr. Inkpen is involved in many collaborative research projects, with University of Toronto, University of Waterloo, IBM Centre for Advanced Studies, and the National Research Council (the Institute for Information Technology and the Canada Institute for Scientific and Technical Information). The team led by Dr. Diana Inkpen and composed of two of her graduate students won the international competition in Information Retrieval CLEF 2005 (Cross-Language Evaluation Forum), the CL-SR task (Cross-Language Spoken Retrieval). |
Date |
Friday September 15, 2006 |
Time |
10:30-12:00 |
Location |
Room 379, Building M-50, NRC Auditorium, 1200 Montreal Road |
Title |
Graph Mining and its Applications |
Speaker |
Jian Pei, Assistant Professor |
| |
Simon Fraser University |
| |
Computing Science Department |
| |
|
| |
|
| |
|
Abstract
Graph models are popularly used in many applications, such as marketing analysis, protein interactions, social networks, and web analysis. Mining significant and interesting graph patterns from collections of graphs as well as other types of data has become an important research problem. In this talk, I shall discuss the problem of mining graph databases and graph patterns in three aspects: how to model patterns in graphs, how to mine large graphs and how to handle many graphs. Particularly, I shall present several interesting approaches recently developed by us. The quasi-clique mining method finds dense areas across multiple large graphs. The ADI approach indexes databases with a large number of (relatively small) graphs and mines frequent sub-graphs. The frequent closed partial order mining approach derives DAG models from large sequence databases. I shall also address the potential extensions of the above methods. |
| |
Speaker Bio
Jian Pei received a Ph.D. degree in Computing Science from Simon Fraser University, in 2002. He is currently an Assistant Professor of Computing Science at Simon Fraser University. His research interests can be summarized as developing effective and efficient data analysis techniques for novel data intensive applications. Particularly, he is currently interested in various techniques of data mining, data warehousing, online analytical processing, and database systems, as well as their applications in bioinformatics, privacy preservation, and education. His current research is supported in part by the Natural Sciences and Engineering Research Council of Canada (NSERC), the National Science Foundation (NSF) of the United States, Hewlett-Packard Company (HP), and the Canadian Imperial Bank of Commerce (CIBC). He has published prolifically in refereed journals, conferences, and workshops, has served extensively in the organization committees and the program committees of many international conferences and workshops, and has been a reviewer for the leading academic journals in his fields. He is a member of the ACM, the ACM SIGMOD, and the ACM SIGKDD. |
Date |
Thursday February 09, 2006 |
Time |
15:00-16:30 |
Location |
NRC Institute for Biological Sciences, 1200 Montreal Road Building M-54, Room 235 |
Title |
A Unified Model of Spike-Time-Dependent Plasticity and Chemotropic Gradients in Retinotopic Map Formation |
Speaker |
Jean-Philippe Thivierge , Psychology Ph.D. candidate |
| |
McGill University |
| |
Laboratory for Natural and Simulated Cognition |
| |
|
| |
|
Abstract
Both activity-dependent and independent processes play a role in the development of the vertebrate visual system. Molecular guidance cues provide a rough topography of early projections, while refinement of termination zones (TZs) occurs later on through correlated retinal activity. Experiments involving B2 subunit knock-out mice have found a cumulative role of removing activity-dependent and independent processes, thus arguing for their distinct roles. A computational account of these results is proposed, based on a unified model that combines chemotropic gradients and spike-time-dependent synaptic plasticity. The model is employed to simulate recent empirical data, and proposes possible interactions between activity-dependent and independent processes. |
| |
Speaker Bio
Jean-Philippe Thivierge is a post-doctoral fellow at the Université de Montréal, where his research interests include the application of computational intelligence tools to cognitive modelling, structures and algorithms in neural networks, bioinformatics, and developmental computational neurobiology. He already has a wide range of publications in these areas, including some very interesting computational models for the development of the visual system. JP has been very active as a young leader in his area of research, in business ventures, and with his professional associations. He was Local Arrangements Chair for the 2005 International Joint Conference on Neural Networks, for which he also chaired a session of international leaders to celebrate developments arising from Donald Hebb's work in Montreal 50 years ago. He also has been a guest editor for the "IEEE International Journal of Neural Networks", "Journal of Machine Learning", and he has been collaborating with the NRC. |
Date |
Thursday December 01, 2005 |
Time |
15:00-16:30 |
Location |
NRC Auditorium, 1200 Montreal Road Building M-50 |
Title |
Computer Vision for Augmented Reality - the ARTag system |
Speaker |
Mark Fiala, Research Officer |
| |
NRC-CNRC Institute for Information Technology |
| |
Computational Video Group |
| |
|
| |
|
Abstract
Augmented Reality (AR) is the convergence of the real world and virtual computer generated imagery, it is the fusion of real and virtual reality through overlaying virtual objects over real images or video. A virtual object can be made to look like it belongs in a real scene if it is rendered from the right viewpoint, something done routinely in movie making but still a research topic for real time systems where you can look at and walk around virtual objects using a head-mounted display, PDA, cellphone, or tablet PC. To do this, the graphics rendering system must know the pose of the camera, this pose determination can be done accurately and inexpensively using computer vision. One way is to use markers like the ARTag marker system that will be described in the talk. Designing markers to add to the environment for robust detection in camera and video imagery is a computer vision application useful to
situations where a camera-object pose is desired such as AR, industrial position tracking, photo-modeling and robot navigation. Examples of augmented reality and the ARTag system developed at the NRC will be shown. |
| |
Speaker Bio
Dr. Mark Fiala is a computer vision researcher at Canada's National Research Council (NRC), where he works in the Computational Video Group centered in Ottawa, Ontario. His work includes fiducial marker systems, panoramic vision, and general computer vision topics such as image segmentation and camera calibration. He graduated from his PhD in Electrical Engineering in 2002 in the field of panoramic computer vision. He also holds an Electrical Engineering BSc and has spent over 5 years in industry in hardware design for imaging and telecom applications. His best known recent work is the "ARTag" fiducial marker system. |
Date
|
Thursday September 29, 2005
|
Time
|
17:00-18:30
|
Location
|
Room 5084 of SITE Building at the University of Ottawa
|
Title
|
|
Speaker
|
Adrian Chan, Assistant Professor
|
| |
Department of Systems and Computer Engineering
|
| |
Carleton University
|
| |
|
| |
|
Abstract
For decades there has been extensive research and
development on myoelectrically controlled powered prostheses.
The basic premise of the device is that myoelectric signals from
residual muscles could be used as a control signal for upper arm
prostheses. The advantages of such a prosthesis are: 1) it frees
the user from straps and harnesses required of body powered and
mechanical switch control; 2) the myoelectric signal can be noninvasively
detected on the surface of the skin; 3) proportional control can
be implemented with relative ease and the amplitude of the myoelectric
signal varies in sympathy with the contractile force; and 4) the
muscle activity required to provide a control signal is relatively
small and can resemble the effort of an intact limb. In recent,
years pattern recognition techniques have been explored to provide
users an interface that is more natural and intuitive, while providing
a higher degree of classification accuracy and controllability.
Work has been extended from this application towards myoelectric
speech recognition; using myoelectric signals from facial muscles
to perform speech recognition. Such a device would be useful for
persons with temporary or permanent speech impairments.
|
| |
Speaker Bio
Dr. Adrian D.C. Chan graduated with his B.A.Sc.
in Computer Engineering, University of Waterloo (1997), M.A.Sc.
in Electrical Engineering, University of Toronto (1999), and Ph.D.
in Electrical Engineering, University of New Brunswick (2002).
Currently, he is an Assistant Professor in the Department of Systems
and Computer Engineering, Carleton University. His research is
in biomedical engineering, focusing on biological signal processing
and noninvasive sensors. Dr. Chan has been recognized as one of
Macleans 25 Best and Brightest (2004) and Ottawa Life Magazine's
Top 50 People in the Capital (2005), and received the Ottawa Life
Sciences Council Dr. Michael Smith Promising Scientist Award (2004).
|
Date
|
Thursday April 21, 2005
|
Time
|
17:00-18:30
|
Location
|
Room 5084 of SITE Building at the University of Ottawa
|
Title
|
|
Speaker
|
Wail Gueaieb, Assistant Professor
|
| |
School of Information Technology and Engineering
|
| |
University of Ottawa
|
| |
|
| |
|
Abstract
A decentralized adaptive fuzzy controller is proposed
for addressing the problem of controlling the positions and internal
forces within multiple coordinated manipulator systems in the
face of parametric and modeling uncertainties as well as external
disturbances. The controller makes use of a multi-input multi-output
fuzzy logic engine and a systematic online adaptation mechanism
to fully approximate the overall system's dynamics. Unlike conventional
adaptive controllers, the proposed controller does not require
a perfect prior model of the system's dynamics nor does it require
a linear parameterization of the system's uncertain physical parameters.
Using a Lyapunov stability approach, the controller is proven
to be robust in the face of varying intensity levels of the aforementioned
uncertainties, and the position and the internal forces are proven
to asymptotically converge to zero under such conditions. Through
a computer simulation of two 3-DOF manipulators, the performance
of the controller is verified and compared to that of one of the
most efficient conventional adaptive controllers proposed in the
literature.
|
| |
Speaker Bio
Dr. Wail Gueaieb received the Bachelor and Master’s
degrees in Computer Engineering and Information Science from Bilkent
University, Turkey, in 1995 and 1997, respectively, and the Ph.D.
in Intelligent Mechatronics from the University of Waterloo, Canada,
in 2001. He then joined Intelligent Mechatronic Systems Inc. in
2001 where he held the positions of a senior systems design engineer
in expert systems and a software manager. During his career at
Intelligent Mechatronic Systems Inc., he worked on the design,
implementation, and productization of a new generation of smart
advanced automotive safety systems. He is also the author/co-author
of three patents. In July 2004, he joined the School of Information
Technology and Information Science (SITE). His areas of expertise
span the fields of intelligent systems design using tools of computational
intelligence with application to a wide range of industries.
|
Date
|
Wednesday February 23, 2005
|
Time
|
18:00-19:30
|
Location
|
Room 5084 of SITE Building at the University of Ottawa
|
Title
|
|
Speaker
|
B. John Oommen, FIEEE, Professor
|
| |
School of Computer Science
|
| |
Carleton University
|
| |
|
| |
|
Abstract
All of the research that has been done in learning,
has involved learning from a Teacher who is either deterministic
or stochastic. In this talk, we shall present the first known
results of how a learning mechanism can learn while interacting
with either a stochastic teacher or a stochastic compulsive
liar. In the first instance, the teacher intends to teach
the learning mechanism. In the second, the compulsive liar intends
to consciously mislead the learning mechanism. We shall present
a formal strategy for the mechanism to perform ε-optimal
learning without it knowing whether it is interacting with a teacher
or a compulsive liar. Believe It Or Not - IT WORKS !
A joint work with Dr. Govindachari (presently in
Bangalore) and Dr. Kuipers (Texas).
|
| |
Speaker Bio
Dr. John Oommen was born in Coonoor, India on September
9, 1953. He obtained his B.Tech. degree from the Indian Institute
of Technology, Madras, India in 1975. He obtained his M.E. from
the Indian Institute of Science in Bangalore, India in 1977. He
then went on for his M.S. and Ph. D. which he obtained from Purdue
University, in West Lafayettte, Indiana in 1979 and 1982 respectively.
He joined the School of Computer Science at Carleton University
in Ottawa, Canada, in the 1981-82 academic year. He is still at
Carleton and holds the rank of a Full Professor. His research
interests include Automata Learning, Adaptive Data Structures,
Statistical and Syntactic Pattern Recognition, Stochastic Algorithms
and Partitioning Algorithms. He is the author of more than 215
refereed journal and conference publications and is a Fellow of
the IEEE. Dr. Oommen is on the Editorial Board of the IEEE
Transactions on Systems, Man and Cybernetics, and Pattern
Recognition.
|
Date
|
Wednesday December 15, 2004
|
Time
|
16:00-17:30
|
Location
|
Room 5084 of SITE Building at the University of Ottawa
|
Title
|
|
Speaker
|
Dmitry O. Gorodnichy, Research Officer
|
| |
NRC-CNRC Institute for Information Technology
|
| |
Computational Video Group
|
| |
|
| |
|
| |
|
The talk will be followed by the annual meeting of the chapter
|
Abstract
The purpose of this talk is two-fold. First, the
audience will be introduced with the basics of the attractor-based
associative neural networks. These networks are known as a mathematical
tool for building recognition systems which work in a fashion
similar to that of the human brain. Second, the audience will
be presented with a new framework for recognizing faces in video.
While the problem of recognizing faces in video has received a
lot of attention recently, in particular, because of such a highly
demanded application of it as security surveillance, this problem
is often erroneously treated as an extension of the problem of
recognizing faces in photographs. Photographs, which are usually
taken under very constrained conditions, provide hard biometrics
data, as do, for instance, the fingerprints. Video footage, on
the other hand, such as the one taken by a surveillance camera,
will very unlikely contain the facial data of high quality and
is therefore the source of "softer" biometrics. As we will show,
however, the soft biometrics provided by video is still very informative
and can be efficiently used to memorize and recognize faces. In
the demonstrations to be shown, the developed mini brain model
allows one to discriminate guests of a talk show in a prerecorded
low-resolution video.
|
| |
Speaker Bio
Dr. Dmitry Gorodnichy is a research officer with
the Computational Video Group of the Institute for Information
Technology of the National Research Council of Canada. He has
two Ph.D. degrees: one - in Computing Science (2000) from the
University of Alberta, Edmonton, Canada, for his work on Vision-based
World Model Learning, and the other (1997) - in Mathematics from
the Glushkov Cybernetics Center of Ukrainian Ac.Sc., Kiev, Ukraine,
for his work on Mathematical models of human memory. His MSc (with
honours) in Information Technology (1994) is from the Moscow Institute
of Physics and Technology, Moscow, Russia. He is the author of
two patents and over thirty conference and journal papers, including
an IJCNN Best Presentation Award paper, a recipient of several
scientific awards, including the Young Investigator Award from
the Canadian Image Processing and Pattern Recognition Society
and the NRC-CNRC Outstanding Scientific Achievement Award. He
is the principle investigator of Nouse™ (Nose as Mouse) and Blink
Detection perceptual vision technologies featured in the 2002
and 2003 NRC-CNRC Annual Reports, and is listed as one of 2003
Leaders of Tomorrow by the Partnership Group for Science and Engineering
of Canada. He was the Program Chair for the International Conference
on Vision Interface, the organizer and the program chair of the
First IEEE Workshop on Face Processing in Video and is now the
Exhibits Chair for the INNS-IEEE International Joint Conference
on Neural Networks to be held in Montreal next year. He is a reviewer
for many scientific conferences, journals and organizations, including
NSERC, and is also presently the Chair for IEEE Computational
Intelligence Society, Ottawa Chapter.
|
Date
|
Thursday November 25, 2004
|
Time
|
16:00-17:00
|
Location
|
Room 5084 of SITE Building at the University of Ottawa
|
Title
|
The heterogeneous neuron model and its use in hybrid neural
networks within computational intelligence compound systems
|
Speaker
|
Julio J. Valdés, Senior Research Officer
|
| |
NRC-CNRC Institute for Information Technology
|
| |
Integrated Reasoning Group
|
| |
|
| |
|
Abstract
A framework is presented for processing heterogeneous
information based on the construction of general observational
domains, and similarity-based function calculi suitable for data
mining and other tasks in domains which can be described by the
corresponding observational models. These calculi are intuitive,
simple, and sufficiently general for classification and pattern
recognition tasks. Functions in these calculi are represented
by a particular kind of neuron models and their behavior is illustrated
with examples from real-world domains showing their capabilities
in processing heterogeneous, incomplete and fuzzy information,
possibly with time dependencies.
|
| |
Speaker Bio
Dr. Julio Valdés has a PhD in mathematics (1987).
His areas of interest are: artificial intelligence (mathematical
foundations of uncertainty processing and inexact reasoning, knowledge
engineering, expert systems and machine learning), digital image
and signal processing, pattern recognition, virtual reality, soft
computing (fuzzy logic, neural networks, evolutionary algorithms,
probabilistic reasoning, rough sets), data mining, data analysis
in general and hybrid systems. He also graduated in geophysics
(1977), oriented to geomathematics, mathematical modeling of natural
processes, computer elaboration and data analysis-mining of earth
science and environmental data, remote sensing, physics and chemistry
of external geodynamic processes and geophysical-geochemical prospecting.
|
Date
|
Wednesday May 05, 2004
|
Time
|
17:00-18:30
|
Location
|
Room 5084 of SITE Building at the University of Ottawa
|
Title
|
Neural Network Modeling of 3D Objects for Virtualized Reality
Applications
|
Speaker
|
Ana-Maria Cretu, Ph.D. Candidate
|
| |
Sensing and Modeling Research Laboratory (SMRLab)
|
| |
School of Information Technology and Engineering
|
| |
University of Ottawa
|
| |
|
| |
|
Abstract
This talk presents a critical comparison between
three neural architectures for 3D object representation in terms
of purpose, computational cost, complexity, conformance and convenience,
ease of manipulation and potential uses in the context of virtualized
reality. Starting from a pointcloud that embeds the shape of the
object to be modeled, a volumetric representation is obtained
using a multilayered feedforward neural network or a surface representation
using either the self-organizing map or the neural gas network.
The representation provided by the neural networks is simple,
compact and accurate. The models can be easily transformed in
size, position (affine transformations) and shape (deformation).
Some potential uses of the presented architectures in the context
of virtualized reality are for the modeling of set operations
and object morphing, for the detection of objects collision and
for object recognition, object motion estimation and segmentation.
|
| |
Speaker Bio
Ana-Maria Cretu obtained her Master degree from
the School of Information Technology and Engineering at the University
of Ottawa, Canada, where she is now a PhD student. Ms. Cretu's
research interests include neural networks, 3D object modeling,
tactile sensing and multi-sensor data fusion. She is a Student
Member of IEEE.
|
|
|
Wednesday February 25, 2004
|
Time
|
16:00-17:00
|
Location
|
Room 5084 of SITE Building at the University of Ottawa
|
Title
|
|
Speaker
|
Peter Turney, Senior Research Officer
|
| |
Information Analysis and Retrieval Group
|
| |
NRC Institute for Information Technology
|
| |
|
| |
|
Abstract
This talk will present an algorithm for learning
from unlabeled text, based on the Vector Space Model (VSM) of
information retrieval, that can solve verbal analogy questions
of the kind found in the Scholastic Aptitude Test (SAT). A verbal
analogy has the form A:B::C:D, meaning "A is to B as C is to D";
for example, mason:stone::carpenter:wood. SAT analogy questions
provide a word pair, A:B, and the problem is to select the most
analogous word pair, C:D, from a set of five choices. The VSM
algorithm correctly answers 47% of a collection of 374 college-level
analogy questions (random guessing would yield 20% correct). This
research is motivated by relating it to work in cognitive science
and linguistics, and by applying it to a difficult problem in
natural language processing, determining semantic relations in
noun-modifier pairs. The problem is to classify a noun-modifier
pair, such as "laser printer", according to the semantic relation
between the noun (printer) and the modifier (laser). The approach
is to use a supervised nearest-neighbour algorithm that assigns
a class to a given noun-modifier pair by finding the most analogous
noun-modifier pair in the training data. With 30 classes of semantic
relations, on a collection of 600 labeled noun-modifier pairs,
the learning algorithm attains an F value of 26.5% (random guessing:
3.3%). With 5 classes of semantic relations, the F value is 43.2%
(random: 20%). The performance is state-of-the-art for these challenging
problems.
|
| |
Speaker Bio
Dr. Peter Turney is a Senior Research Officer in
the Interactive Information Group of the National Research Council.
In 1988, he obtained his PhD from the University of Toronto, where
he then accepted a Postdoctoral Fellowship. He joined the NRC
in 1989, and he has since worked on a variety of projects, all
involving applications of machine learning technology. His recent
work focuses on machine learning applied to natural language.
He is the author or co-author of more than sixty publications,
a past editor of Canadian Artificial Intelligence magazine, and
a member of the Advisory Board of the Journal of Artificial Intelligence
Research.
|
|
|
Wednesday December 03, 2003
|
Time
|
16:00-17:30
|
Location
|
Room 5084 of SITE Building at the University of Ottawa
|
Title
|
Hardware Neural Network Architectures Using Random Data Representation
|
Speaker
|
Emil M. Petriu, Dr. Eng., P.Eng., Professor
|
| |
Sensing and Modeling Research Laboratory (SMRLab)
|
| |
School of Information Technology and Engineering
|
| |
University of Ottawa
|
| |
|
| |
|
Abstract
The idea of using computational techniques that
mimic the processing behaviour of the biological nervous systems
was advanced by von Neuman in 1965. The resulting random-pulse
machine concept deals with analog variables represented by the
mean rate of random-pulse streams using simple digital circuits
to perform arithmetic and logic operations. This concept presents
a good trade-off between the electronic circuit complexity and
the computational accuracy.
The talk presents the random-pulse data representation
and discusses how it can be used to the design of modular random-pulse
neural networks. A generalization of the random-pulse machine
concept, namely the multi-bit random-data machine, is then discussed.
The advantage of using multi-bit data instead of pulses (which
are 1-bit data) is a considerable reduction in the time needed
to get an acceptable accuracy for the statistical averages of
the data streams carrying the information. As in the case of the
random-pulse machine, the arithmetic operations are performed
by relatively simple logic circuits. The resulting architectures
have high functional packing density making them suitable for
the VLSI implementation of hardware parallel neural networks.
|
| |
Speaker Bio
Emil M. Petriu is a professor in the School of Information
Technology and Engineering at the University of Ottawa, Canada,
where he has been since 1985. Dr. Petriu's research interests
include intelligent sensors, robot sensing and perception, neural
networks, and fuzzy control. During his career he has published
more than 180 technical papers, authored two books, edited other
two books, and received two patents. He is a Fellow of IEEE, Fellow
of the Canadian Academy of Engineering, and Fellow of the Engineering
Institute of Canada. For more info, see www.smrlab.uottawa.ca
|
|
|
Friday April 25, 2003
|
Time
|
16:00-17:00
|
Location
|
Room 5084 of SITE Building at the University of Ottawa
|
Title
|
Collaborative Virtual Environments
|
Speaker
|
Nicolas D. Georganas, FIEEE, Professor
|
| |
Distributed and Collaborative Virtual Environments Research
Laboratory (DISCOVER)
|
| |
School of Information Technology and Engineering
|
| |
University of Ottawa
|
| |
|
Abstract
One of the hottest topics in Virtual Reality research
is that of "Distributed" Virtual Environments (DVE). The idea
behind DVE is very simple; a simulated world runs not on one computer
system, but on several, using a series of client server applications.
The computers are connected over a network and people using those
computers are able to interact in real time, sharing the same
virtual world.
Collaborative Virtual Environments (CVE) add new
dimensions to human-factors, networking, and database issues.
For example, human-factors research in VR has traditionally focused
on the development of natural interfaces for manipulating virtual
objects and traversing virtual landscapes. Collaborative manipulation,
on the other hand, requires the consideration of how participants
should interact with each other in a shared space, in addition
to how co-manipulated objects should behave. Other issues include:
how participants should be represented in the collaborative environment;
how to effectively transmit non-verbal cues that real-world collaborators
so casually and effectively use; how to best transmit video and
audio via a channel that allows both public addressing as well
as private conversations to occur; how to filter relevant information
to reduce processing (increase performance) at each client for
large worlds; and how to sustain a virtual environment even when
all its participants have left.
This talk will expose basic notions in CVE and describe
several applications.
|
| |
Speaker Bio
Nicolas D. Georganas, is Distinguished University
Professor and Canada Research Chair in Information Technology,
School of Information Technology and Engineering, University of
Ottawa, Canada. He is a Fellow of IEEE, Fellow of the Canadian
Academy of Engineering, Fellow of the Engineering Institute of
Canada, and Fellow of the Royal Society of Canada. In 2002, he
received the Killam Prize for Engineering, Canada's highest award
for career achievements in research. His research interests are
in Multimedia Communications, Pervasive Computing, Intelligent
Sensors, Tele-Haptics and Collaborative Virtual Environments.
For more info, see www.discover.uottawa.ca
|
|
