The talk provides an insight into what has been achieved in the past, what is currently hot and what would eventually be the future trend in the area of machine learning and pattern recognition.
The speech is divided into two parts that contain: (i) key dimensionality reduction (DR) techniques and (ii) some major classification paradigms. We will cover classical linear DR techniques and newly developed nonlinear ones such as those exhibiting a mapping (either from the high-dimensional space to the low-dimensional embedding or vice versa), and those that just provide a visualization.
The second part focuses on new perspectives and achievements in classification such as the concepts of “deep learning” and “convolutional neural networks”, as well as ensemble classification such as“Common Weighting Methods”(performance weighting, majority voting) and “Common Meta-combination Methods” (meta-learning, stacking). We conclude the talk with some code demonstrations on selected data related to the materials covered.
Prof. Djamel Bouchaffra is a Director of Research at the Center for Development of Advanced Technologies. He is currently the Head of the ASM Division. Prior to this appointment, Dr. Djamel Bouchaffra was a Professor of Computer Science at Grambling State University. He started hisacademic position at Joseph Fourier University, Grenoble (France), and then he held a position of Assistant Professor at Oakland University, Michigan(USA). Professor Bouchaffra was teaching a wide spectrum of courses including machine learning and pattern recognition, artificial intelligence,soft computing and discrete mathematics. Dr. Bouchaffra has been selected as the recipient of the Oakland University Teaching Excellence Award for2004(awardees) as well as the recipient of the School of Engineering Teaching Excellence Award. This recognition acknowledges the genuine admiration of his students and his colleagues for his superior level of professionalism in teaching. His personal views on the art of teaching are published in thisOakland University Newsletter link. In February 2005, he was invited by OU-TV “Focus on FacultyProgram” broadcasted on Comcast cable channel 74. Prior to this appointment, he held a position ofSenior Research Scientist at the “Center of Excellence for Document Analysis and Recognition”(CEDAR) located at The State University of New York. He was a technical lead (research director) of different research groups in several federally sponsored projects.
His field of research is in Pattern Recognition and Machine Learning, Computational Intelligence, Computer Vision, and Signal Processing. He is currently working on the approximation and the learning of topological manifolds embedded in a high-dimensional vector space for object recognition. Professor Bouchaffra introduced both the structural and the topological hidden Markov models as two paradigms that implement this fusion. His areas of applications include (but are not limited to!) Brain Machine Interface, Remote Sensing, Biometrics, Handwriting Recognition, LanguageModeling, Speech Recognition, and Bio-informatics. He is the director of the “Design andImplementation of Intelligent Machines” (DIIM).
He has written several papers in peer-reviewed conferences and premier journals. He chaired several sessions in conferences, he was one of the general chairs of the IEEE conference ICSIT'05(http://www.cdta.dz/sitedasm/manifestations/icsit05/executive_committee.htm) held in Algiers. He also served as a Vice Chair at the 9th International Workshop on Systems, Signal Processing and theirApplications (WoSSPA'13)(http://www.cdta.dz/manifestations/wosspa2013/contenu/committees/organizing-committee-.html . Prof. Bouchaffra is among the reviewer panel of some governmental fundiing agencies such as NASA (http://www.nasa.gov/) (ADP Program: Data Analysis and Astrophysics) and(EPSRC): (http://www.epsrc.ac.uk/default.htsm). He is also a regular reviewer for several journals such as IEEE TPAMI, TNN, TKDE, and Image Processing. He is a currently the lead guest editor of as pecial issue of Pattern Recognition journal (Elsevier) titled: “Machine Learning and PatternRecognition Models for Change Detection”. He was also the lead guest Editor for Pattern Recognition journal titled: "Feature Generation and Machine Learning for Robust Multimodal Biometrics", volume41, number 3 that has been published in March 2008 as well as Advances in Artificial IntelligenceJournal titled: "Machine Learning Paradigms for Modeling Spatial and Temporal Information inMultimedia Data Mining"). He has been an “invited Professor” by University Paris 13 (Laboratoired'Informatique de Paris-Nord UMR CNRS 7030) for the summer 2010 and 2011.
Professor Bouchaffra is an editorial board member of the “Journal of Pattern Recognition”(published by Elsevier), “The Open Information Systems Journal”, Bentham Science Publisher, an associate editor for the Journal titled “Advances in Artificial Intelligence” (published by HindawiCorporation), “the Journal of Engineering Letters” (published by the International Association ofEngineers), the “Scientific Journals International” (SJI) and Transactions on Machine Learning andArtificial Intelligence (Society for Science and Education, UK) .
The Institute of Electrical and Electronic Engineering at the University of Boumerdes formerly called the National Institute of Electricity and Electronics- INELEC, came into existence as a result of an agreement signed between the National Company for Manufacture and Assembly of Electrical and Electronic Equipment, SONELEC, and Education Development Center, EDC, Newton, Mass., USA. The agreement was signed as early as March 1976, according to the program set forth in the planning study conducted for SONATRACH in March 1974.
... read more