Seminario “La IA en la adquisición y uso del conocimiento médico: Patrones Asistenciales”

El Jueves 21 de Mayo de 2009 se celebró el seminario “La IA en la adquisición y uso del conocimiento médico: Patrones Asistenciales” en la ETSI Informática de la UNED. El seminario presentaba el proyecto HYGIA (TIN2006-15453), el cual integra el trabajo conjunto desarrollado por las universidades de Santiago de Compostela, Jaume I y Rovira i Virgili, y del Hospital Clínico de Barcelona en el ámbito de la utilización de técnicas de análisis automático de textos, ingeniería del conocimiento y aprendizaje automático inductivo para la generación de patrones asistenciales. Estos patrones formalizan el conocimiento terapéutico de una o varias enfermedades y son capaces de ser interpretados por sistemas informáticos de soporte a la toma de decisiones en medicina. En esta ponencia se presentó el trabajo y los resultados obtenidos en el análisis automático de textos médicos y en el aprendizaje inductivo de patrones a partir de las bases de datos hospitalarias.
La web oficial del proyecto es: http://banzai-deim.urv.net/~riano/TIN2006-15453/

Presentation of the multikulti algorithm at CEC09

The Congress of Evolutionary Computation is one of the bigger and more prestigious conferences on, you guessed it, EC. Here are the slides for the talk Lourdes Araújo and myself will be presenting tomorrow.

The paper is available under request, but I guess it will be available shortly from IEEExplore. Here’s the abstract

Migration policies in distributed evolutionary algorithms are bound to have, as much as any other evolutionary operator, an impact on the overall performance. However, they have not been an active area of research until recently, and this research has concentrated on the migration rate. In this paper we compare different migration policies, including our proposed ‘multikulti’ methods, which choose the individuals that are going to be sent to other nodes based on the principle of ‘multiculturalism’: the individual sent should be as different as possible to the receiving population (represented in several possible ways). We have checked this policy on two discrete optimization problems for different number of nodes, and found that, in average or in median, multikulti policies outperform others like sending the best or a random individual; however, their advantage changes with the number of nodes involved and the difficulty of the problem. The success of this kind of policies will be explained via the measurement of entropies, which are known to have an impact in the performance of the evolutionary algorithm.