Concordia University

Artificial Neural Networks in Pattern Recognition ANNPR 2014

October 6-8, 2014, Concordia University, Montreal, Canada

Conference program

All sessions are oral presentations and will take place on the second floor (EV 2.260 and EV 2.184) in Concordia University EV building located at 1515 St. Catherine St. West above Guy metro station (green line).

DAY 1 - Monday October 6, 2014
8.00 - 9.00 Registration
9.00 - 9.15 Opening & Welcome Speech
9.15 -10.15

IAPR Invited Speaker: Zhi-Hua Zhou
Session Chair: Dr. Ching Suen
Title: Large Margin Distribution Learning

Abstract: Support vector machines (SVMs) and Boosting are possibly the two most popular learning approaches during the past two decades. It is well known that the margin is a fundamental issue of SVMs, whereas recently the margin theory for Boosting has been defended, establishing a connection between these two mainstream approaches. The recent theoretical results disclosed that the margin distribution rather than a single margin is really crucial for the generalization performance, and suggested to optimize the margin distribution by maximizing the margin mean and minimizing the margin variance simultaneously. Inspired by this recognition, we advocate the large margin distribution learning, a promising research direction that has exhibited superiority in algorithm designs to traditional large margin learning.

10.15 - 10.45 Coffee Break
10.45 - 12.25

Session 1 : Learning and Architectures
Session Chair: Robert Sabourin

A Decorrelation Approach for Pruning of Multilayer Perceptron Networks
Hazem M. Abbas

Combining Bipartite Graph Matching and Beam Search for Graph Edit Distance Approximation
Kaspar Riesen, Andreas Fischer, Horst Bunke

Hidden Markov Models Based on Generalized Dirichlet Mixtures for Proportional Data Modeling
Elise Epaillard, Nizar Bouguila

Linear Contrast Classifiers in High-Dimensional Spaces
Florian Schmid, Ludwig Lausser, Hans A. Kestler

12.30 - 14.00 Lunch
14.00 - 15.15 Session 2 : Ensembles
Session Chair: Hazem Abbas

Dynamic Weighted Fusion of Adaptive Classifier Ensembles Based on Changing Data Streams
Christophe Pagano, Eric Granger, Robert Sabourin, Gian Luca Marcialis, Fabio Roli

Forward and Backward Forecasting Ensembles for the Estimation of Time Series Missing Data
Tawfik A. Mohamed, Neamat El Gayar, Amir F. Atiya

Analyzing Dynamic Ensemble Selection Techniques Using Dissimilarity Analysis
Rafael M. O. Cruz, Robert Sabourin, George D. C. Cavalcanti
15.15 - 15.45 Coffee Break
15.45 - 17.00 Session 3: Ensembles & Applications
Session Chair: George Cavalcanti

Intelligent Ensemble Systems for Modeling NASDAQ Microstructure: A Comparative Study
Salim Lahmiri

Face Recognition Based on Discriminative Dictionary with Multilevel Feature Fusion
Hongjun Li, Nicola Nobile, Ching Y. Suen

A Time Series Classification Approach for Motion Analysis Using Ensembles in Ubiquitous Healthcare Systems
Rana Salaheldin, Mohamed ElHelw, Neamat El Gayar
18.00 - 20.00 Welcome reception at Concordia University, room EV 11.725
DAY 2 - Tuesday, October 7, 2014
9.00 - 10.00

IAPR Invited Speaker: Yoshua Bengio
Session Chair: Friedhelm Schwenker
Title: Deep Learning

Abstract: Deep learning has arisen around 2006 as a renewal of neural networks research allowing such models to have more layers. Theoretical investigations have shown that functions obtained as deep compositions of simpler functions can express highly varying functions (with many ups and downs and different input regions that can be distinguished) much more efficiently (with fewer parameters) than otherwise. Empirical work in a variety of applications has demonstrated that, when well trained, such deep architectures can be highly successful, remarkably breaking through previous state-of-the-art in many areas, including speech recognition, object recognition, language models, and transfer learning. This talk will summarize the advances that have made these breakthroughs possible, and end with questions about some major challenges still ahead of researchers in order to continue our climb towards AI-level competence.

10.00 - 10.30 Coffee Break
10.30 - 12.10 Session 1 : Support Vector Machines & Applications
Session Chair: Adam Krzyzak

A New Multi-Class Fuzzy Support Vector Machine Algorithm
Friedhelm Schwenker, Markus Frey, Michael Glodek, Markus Kächele, Sascha Meudt, Martin Schels and Miriam Schmidt

Incremental Feature Selection by Block Addition and Block Deletion Using Least Squares SVRs
Shigeo Abe

Majority-Class Aware Support Vector Domain Oversampling for Imbalanced Classification Problems
Markus Kächele, Patrick Thiam, Günther Palm and Friedhelm Schwenker

Prediction of Insertion-Site Preferences of Transposons Using Support Vector Machines and Artificial Neural Networks
Maryam Ayat, Michael Domaratzki
12.10 - 14.00 Lunch
14.00 - 15.15 Session 2: Character Recognition
Session Chair: Andreas Fischer

End-Shape Recognition for Arabic Handwritten Text Segmentation
Amani T. Jamal, Nicola Nobile, Ching Y. Suen

Part-Based High Accuracy Recognition of Serial Numbers in Bank Notes
Bo-Yuan Feng, Mingwu Ren, Xu-Yao Zhang, Ching Y. Suen

Investigating of Preprocessing Techniques and Novel Features in Recognition of Handwritten Arabic Characters
Ahmed T. Sahlol, Ching Y. Suen, Abdelhay A. Sallam, Mohammed R. Elbasyoni
15.15 - 15.45 Coffee Break
15.45 - 17.00 Session 3: Image Processing
Session Chair: Tony Kasvand

Automatic Bridge Crack Detection { A Texture Analysis-Based Approach }
Sukalpa Chanda, Guoping Bu, Hong Guan, Jun Jo, Umapada Pal, Yew-Chaye Loo, Michael Blumenstein

Comparative Study of Feature Selection for White Blood Cell Differential Counts in Low Resolution Images
Mehdi Habibzadeh, Adam Krzyzak, Thomas Fevens

Low-dimensional Data Representation in Data Analysis
Alexander Bernstein, Alexander Kuleshov
18.00 - 21.00 Dinner at the Vieux-Port Steakhouse in old Montreal
(39 Rue Saint Paul Est, Montréal, QC H2Y 1G2, Canada)
DAY 3 - Wednesday, October 8, 2014
9.00 - 10.00

IAPR Invited Speaker: J. Michael Herrmann
Session Chair: Neamat El Gayar
Title: Pattern Formation and Pattern Recognition: Active Learning in Biologically Inspired Robotics

Abstract: The concept of self-organised criticality (SOC) describes the dynamics of systems as diverse as sandpiles, earthquakes, evolutionary systems, animal swarms, financial markets and the brain. It is particularly interesting in neural networks where SOC has been shown to enable optimal transmission and storage of information, and to improve sensing and control which is particularly useful for applications in autonomous robots. Operation at or near a critical point enables a robot to generate a sequence of optimal training data for learning a model of the environment. This active learning procedure can be used also as an organisational principle for a deep neural control architecture in a robot which serves the double function of generating input patterns by exploratory behaviour and recognising environmental perturbations of the self-generated patterns. The dynamics of such an adaptive control system is analysed in theory, in robotic experiments and discussed as an abstract model of brain-style control of behaviour. Following the idea of guided self-organisation, we show how pattern formation in critical systems can improve learning capabilities if, as often in applications, an external evaluative signal is available.

10.00 - 10.30 Coffee Break
10.30 - 12.10 Session 1: Learning II
Session Chair: Shigeo Abe

Unsupervised Active Learning of CRF Model for Cross-Lingual Named Entity Recognition
Mohamed Farouk Abdel Hady, Abubakrelsedik Karali, Eslam Kamal, Rania Ibrahim

A Reinforcement Learning Algorithm to Train a Tetris Playing Agent
Patrick Thiam, Viktor Kessler, Friedhelm Schwenker

Bio-Inspired Optic ow From Event-Based Neuromorphic Sensor Input
Stephan Tschechne, Roman Sailer, Heiko Neumann

Computing Upper and Lower Bounds of Graph Edit Distance in Cubic Time
Kaspar Riesen, Andreas Fischer, Horst Bunke
12.10 -14.00 Closing Remarks
Back to top

© Concordia University