DTAI

Monday, May 21 2012, at 16h30 in Celestijnenlaan 200A (auditorium 00.225)

A Relational Kernel-based Approach to Scene Classification
by Laura Antanas (PhD student DTAI)

Relational representations have recently been combined with statistical and probabilistic methods and successfully applied to a number of artificial intelligence applications. We show that such representations and techniques can be also useful for computer vision.  More specifically, we employ kLog, a logical and relational language for learning with kernels to improve scene classification. The key advantage of kLog is that both appearance features and rich dependencies between objects in a scene can be integrated in an easy, elegant and interpretable formulation to obtain a semantic representation of the problem domain.  We experiment on real-world datasets and obtain encouraging results by combining appearance cues with symbolic relations in a unified framework.

Wednesday May 2, 2012, at 10h00 in Celestijnenlaan 200A (room 200A.05.001)

Automatic Discovery for Research in Machine Learning
by Dr. Francis Maes

Machine learning research aims at conceiving and analyzing algorithms to solve specific data-related problems. Researchers in this field typically use a trial-and-error approach to reach these goals, in which algorithms are progressively improved on the basis of empirical evidence and/or theoretical analysis. This research process has some important drawbacks: it is inherently biased towards solutions that seem intuitive to humans, it is time-consuming and it is rarely reproducible. In this talk, I will defend the approach which consists in formalizing this trial-and-error approach and in using automatic discovery tools to discover new high-performance algorithms. The talk is structured in two main parts. In the first one, I will formalize the search for high-performance algorithms as a multi-armed bandit (MAB) problem and illustrate the MAB approach on three different problems: discovering control policies, discovering MAB algorithms and discovering reinforcement learning algorithms. In a second time, I will consider the problem of exploring large tree-structured expression spaces and propose some alternatives to Genetic Programming for this task. The proposed approaches formalizes the problem as a "one-player game" and relies on Monte-Carlo search techniques to solve this game. I will illustrate it on symbolic regression and on automatic feature generation. Through these numerous examples, my hope is to show that automatic discovery tools are extremely relevant to research in machine learning and that these tools are a key to quickly unlock several new research avenues.

Pieter Wuille will be defending his doctoral thesis, entitled "Functional programming abstractions for CP modeling."  All are invited.

The defense will take place on December 6th, 2011 in Auditorium De Molen. More information, including a form for confirming your presence, can be found at http://people.cs.kuleuven.be/~pieter.wuille/phd.php

On Friday, December 19, the Declarative Languages and Artificial Intelligence research group of the Katholieke Universiteit Leuven organizes the symposium "Consistency Handling in Relational Databases" at the Computer Science Department of the K.U. Leuven, Celestijnenlaan 200A, 3001 Heverlee, Belgium.

Attending the symposium is free, but registration is required (by December 16).

Program:

13:00 - "Distance Semantics for Reasoning with Inconsistency in Information Systems" by Ofer Arieli (Academic College of Tel-Aviv, Israel)
14:00 - "Consistent Query Answering: The First Ten Years" By Jan Chomicki (University at Buffalo, NY, USA)
15:00 - Coffee break
15:30 - "Consistent query answering under primary key constraints ". By Jef Wijsen Université de Mons-Hainaut, Mons, Belgium
17:00 - PhD Defense Álvaro Cortés: "Towards a Logical Reconstruction of a Theory for Locally Complete Databases"

For more information about the program, registration and location, please visit:http://www.cs.kuleuven.ac.be/~alvaro/phddefense/symposium.html