Four workshops, conducted by each Keynote speaker, are planned to happen on Tuesday, August 30. Those interested must register in the category “Workshop” and contact us, via email or by phone, to complete the process by registering the workshop of your choice. Each one has an additional cost of COP 60.000 (USD 20) and is limited to only 15 participants, so hurry up and register now.

For more information see our registration rates.

Information Extraction from Speech: Humans and Machines

by Hynek Hermansky, PhD

It is easy to argue that speech recognition engineering should apply knowledge of properties of human auditory perception – both have the goal of extracting message from speech. However, hearing research can also learn from advances in speech technology: the knowledge that helps in engineering applications lends support to hearing theories. Further, fundamental difficulties that speech engineering still faces could indicate weakness in our current understanding of the human speech communication process and should guide further inquires into it. The tutorial covers history of processing of speech, and some results from human speech perception, leading to currently dominant techniques for acoustic processing of speech signals in machine recognition of speech. The emphasis is on techniques from our research group, which are motivated by some basic properties of human speech processing.

SHAPE ANALYSIS AND MODELING IN VIDEO APPLICATIONS: ACTIVITY ANALYSIS

by Hamid Krim, PhD

High dimensional data exhibit distinct properties compared to its low dimensional counterpart; this causes a common performance decrease and a formidable computational cost increase of traditional approaches. Novel methodologies are therefore needed to characterize data in high dimensional spaces. Considering the parsimonious degrees of freedom of high dimensional data compared to its dimensionality, we study the union-of-subspaces (UoS) model, as a generalization of the linear subspace model. The UoS model preserves the simplicity of the linear subspace model, and enjoys the additional ability to address nonlinear data. We show a sufficient condition to use l1 minimization to reveal the underlying UoS structure, and further propose a bi-sparsity model (RoSure) as an effective algorithm, to recover the given data characterized by the UoS model from errors/corruptions. As an interesting twist on the related problem of Dictionary Learning Problem, we discuss the sparse null space problem (SNS). Based on linear equality constraint, it first appeared in 1986 and has since inspired results, such as sparse basis pursuit, we investigate its relation to the analysis dictionary learning problem, and show that the SNS problem plays a central role, and may naturally be exploited to solve dictionary learning problems. Substantiating examples are provided, and the application and performance of these approaches are demonstrated on a wide range of problems, such as face clustering and video segmentation.

Probabilistic Graphical Models and their Applications in Vision

by Luis Enrique Sucar Succar, PhD

Probabilistic graphical models provide a general framework for solving complex problems efficiently, by representing the conditional independencies between a set of variables. In this tutorial we will give a general introduction to probabilistic graphical models. We will cover different types of models: Bayesian classifiers, hidden Markov models and Markov random fields; including representation, inference and learning techniques. We will then illustrate the application of these models in several problems in computer vision, such as person detection, gesture recognition and image retrieval.

Introduction to Hyper/Multispectral Remote Sensing

by Miguel Velez Reyes, PhD

This tutorial will provide an overview of multi/hyperspectral remote sensing systems. Introduction to imaging principles for optical systems used in multi/hyperspectral remote sensing. Discussion of existing and proposed airborne and satellite remote sensing platforms. Overview of the end-to-end information processing chain including algorithms, methodologies and tools for information extraction and management in multi/hyperspectral remote sensing. Specific processing examples involving image enhancement, feature extraction and classification, clustering, and multi/hyperspectral unmixing will be discussed. Application examples will be discussed. Examples based on the UPRM/UTEP Hyperspectral Image Processing Toolbox will be presented.