CIARP 2010 Home | News | Accepted Papers | Program | Contact Us             Guadalajara, Mexico

   » Home
   » News
   » Conference Info
   » Invited Speakers
   » Program
   » Accepted Papers
   » Award to the best paper
   » Committees
   » Venue
   » Accommodation
   » São Paulo
   » Registration
   » Org. ,Support & Sponsors
   » Intructions for authors
   » Paper Submission
   » Ciarp 2009




Invited speakers

Seth Hutchinson
Dept. of Electrical and Computer Engineering
The Beckman Institute
University of Illinois  External Link
Talk: Vision-Based Control of Robot Motion

Visual servo control is now a mature method for controlling robots using real-time vision feedback. It can be considered as the fusion of computer vision, robotics and control, and it has been a distinct field since the 1990's, though the earliest work dates back to the 1980's. Over this period several major, and well understood, approaches have evolved and have been demonstrated in many laboratories around the world. Many visual servo schemes can be classified as either position-based or image-based, depending on whether camera pose or image features are used in the control law. This lecture will review both position-based and image-based methods for visual servo control, presenting the basic derivations and concepts, and describing a few of the performance problems faced by each. Following this, a few recent and more advanced methods will be described. These approaches essentially partition the control system either along spatial or temporal dimensions. The former are commonly referred to as hybrid or partitioned control systems, while the latter are typically referred to as switched systems.

Stéphane Canu
Director of Laboratory LITIS
France  External Link Canu
Talk: Recent advances in kernel machines

This talk will review recent advances in the kernel methods focusing on support vector machines (SVM) for pattern recognition. Topics discussed include the kernel design issue through the multi kernel approach and the optimization issue with emphasis on scalability and non convex cost functions.

Tutorial: Introduction to Kernel Machines

Kernel Machines is a term covering a large class of learning algorithms, including Splines and support vector machines (SVM) as a particular instance. Kernel Machines is an important and active field of all Machine Learning research. Not only the number of publications bear witness of this fact but also the high quality of the results obtained by kernel machines in recent pattern recognition competitions. This tutorial will provide an introduction to kernel machines by explaining how and why it works. It will be organized in three parts dealing with the problem: kernels and learning (part 1), tools: kernels, functions, costs and optimization (part 2), and algorithms for non sparse and sparse kernel machines (part 3).

Alexandre X. Falcão
Professor Doctor
Institute of Computing (IC)
University of Campinas (UNICAMP)  External Link
Talk: Design of Pattern Classifiers using Optimum-Path Forest with Applications in Image Analysis

Current image acquisition and storage technologies have provided large data sets (with millions of samples) for analysis. Samples may be images from an image database, objects extracted from several images, or image pixels. This scenario is very challenging for traditional machine learning and pattern recognition techniques, which need to be more efficient and effective in large data sets. This lecture presents a recent and successful methodology, which links training samples in a given feature space and exploits optimum connectivity between them to the design of pattern classifiers. The methodology essentially extends the Image Foresting Transform, successfully used for filtering, segmentation and shape description, from the image domain to the feature space. Several supervised and unsupervised learning techniques may be created from the specification of two parameters: an adjacency relation and a connectivity function. The adjacency relation defines which samples form arcs of a graph in the feature space.

The connectivity function assigns a value to any path in the graph. The path value indicates the strength of connectedness of its terminal node with respect to its source node. A connectivity map is maximized by partitioning the graph into an optimum-path forest rooted at its maxima (i.e., representative samples of each class/group, called prototypes). The optimum-path forest is then a pattern classifier, which assigns to any new sample the class (or group label) of its most strongly connected root. The methods have been successfully applied to several applications and this lecture demonstrates two recent ones: content-based image retrieval (CBIR) and 3D segmentation of brain tissues in MR images. In CBIR, user interaction is considerably reduced to a few clicks on relevant/irrelevant images along 3 iterations of relevance feedback followed by supervised learning in order to achieve satisfactory query results. The 3D segmentation of brain tissues is automatically obtained in less than 2 minutes. It exploits voxel clustering, some prior knowledge and does not require a brain atlas for that purpose, while many other brain tissue segmentation methods do. The lecture concludes by discussing some open problems and perspectives for the optimum-path forest classifiers.

Matthew Turk
Computer Science Department
University of California, Santa Barbara  External Link
Talk: Computational Illumination

The field of computational photography includes computational imaging techniques that enhance or extend the capabilities of digital photography, a combination of computer vision, computer graphics, and applied optics. Computational illumination is an aspect of computational photography that considers how to modify illumination in order to facilitate useful techniques in computer vision and imaging. This talk will present research using multiflash imaging, coded shadow photography, and parameterized structured light, three families of techniques in computational illumination, where the results help to produce reliable information in scenes that is often difficult to robustly compute otherwise.

Tutorial: Multimodal Human-Computer Interaction for Mobile Computing

We are in the midst of a revolution in computing, in which mobile computing devices are displacing traditional computing platforms in importance. State-of-the-art mobile phones have cameras, accelerometers, GPS, magnetometers, microphones, keyboards, and touch-sensitive displays, not to mention ever-increasing computation power and memory, graphics capabilities, and various communications capabilities. Given the significant market penetration and their importance in people's daily lives, there are wonderful opportunities for research and for creative applications that take advantage of this rich computing platform. In this tutorial, we will explore new and recent opportunities for pattern recognition, and especially for multimodal processing, in the context of mobile computing.

Sankar Kumar Pal
Director and Distinguished Scientist
Indian Statistical Institute  External Link Kumar
Tutorial: Soft Computing, f-Granulation and Pattern Recognition

Different components of soft computing (e.g., fuzzy logic, artificial neural networks, rough sets and genetic algorithms) and machine intelligence, and their relevance to pattern recognition and data mining are explained. Characteristic features of these tools are described conceptually. Various ways of integrating these tools for application specific merits are described. Tasks like case (prototype) generation, rule generation, knowledge encoding, classification and clustering are considered in general. Merits of some of these integrations in terms of performance, computation time, network size, uncertainty handling etc. are explained; thereby making them suitable for data mining and knowledge discovery.

Granular computing through rough sets and role of fuzzy granulation (f-granulation) is given emphasis. Different applications of rough granules and certain challenging issues in their implementations are stated. The significance of rough-fuzzy computing, as a stronger paradigm for uncertainty handling, and the role of granules used therein are explained with examples. These include tasks such as class-dependent rough-fuzzy granulation for classification, rough-fuzzy clustering, and defining generalized rough entropy for image ambiguity measures and analysis. Image ambiguity measures take into account the fuzziness in boundary regions, as well as the rough resemblance among nearby gray levels and nearby pixels.

Significance of rough granules and merits of some of the algorithms are described on various real life problems including multi-spectral image segmentation, determining bio-bases (c-medoids) in encoding protein sequence for analysis, and categorizing of web document pages (using vector space model) and web services (using tensor space model). The talk concludes with stating the possible future uses of the methodologies, relation with computational theory of perception (CTP), and the challenges in mining.

 Important Dates

» The conference will be held on November 08-11, 2010

»  Submission of papers: June 22nd, 2010
[ Closed ]

»   Notification of acceptance July 29th, 2010
[ Closed ]

»  Camera-ready August 15th, 2010
[ Closed ]

Second Call For Papers
download Call For Paper
Home | Contact Us                   
Copyright © 2009 - 2010 Advanced Technologies Applications Center of Cuba