Fabio Cozman's - Home Page

Who I am

I am a Full Professor at Escola Politécnica (Poli), Universidade de São Paulo (USP), Director of the Center for Artificial Intelligence at USP, and researcher (level 1C-CNPq/Pq) with an interest in machine learning and knowledge/uncertainty representation.

I have served, amongst other duties, as Program and General Chair of the Conference on Uncertainty in Artificial Intelligence, Area Chair of the International Joint Conference on Artificial Intelligence, and Associate Editor of the Artificial Intelligence Journal, the Journal of Artificial Intelligence Research, and the Journal of Approximate Reasoning. I chaired the Special Committee on Artificial Intelligence of the Brazilian Computer Society, and received the Prize for Scientific Merit in Artificial Intelligence by that society.

I took Engineering at Poli-USP and got a PhD from Carnegie Mellon University, where I worked in the Robotics Institute; my advisor was Eric Krotkov, the Great. During my PhD I was in the VASC group, working with the Lunar Rover project, and more specifically in the Viper system. While at CMU, I used to play soccer with the Robotics team, the Viking Death Rats. I was captain of the team for four years (we were Intramurals champions in 1995). After my PhD I returned to USP where I worked at Escola Politécnica (Engineering School). I have been the Undergraduate Chair of the school, and also I have been head of the Mechatronics Engineering Department.

I now teach classes in Artificial Intelligence, Machine Learning, and Probability Theory.

My Lattes Curriculum Vitae

Research

Artificial Intelligence

I work within the broad area of artificial intelligence, mostly with machine learning and knowledge representation. For my current work you can take a look at the Center for Artificial Intelligence, where most of my activities take place.

Knowledge Representation and Machine Learning

I have been working recently on the interplay between knowledge representation and machine learning. I have been particularly interested in question answering systems, in physics-informed machine learning, in the interpretability of artificial intelligence devices, and in their social impact.

Probabilistic Reasoning

I have worked with probabilistic reasoning, both with machine learning techniques and with the representation of facts, rules and beliefs. I have investigated languages that specify knowledge and beliefs, often inspired by Bayesian networks; for instance, relational Bayesian networks and probabilistic logic programs. I have studied their complexity and expressivity, and developed algorithms for inference and learning with them. Years ago I coded the JavaBayes package for Bayesian networks, and was involved in the Interchange Format for Bayesian networks.

Theory of Probabilities

I have worked with various alternatives to the theory of probabilities, for instance dealing with full conditional probabilities or credal sets (sets of probabilities). You can take a look at the web site of the Society for Imprecise Probability Theory and Applications (SIPTA); I'm a founding member of this society and I was the society's president for a while. I have examined concepts of independence and decision making strategies; developed graph-based models usually called credal networks; developed algorithms that compute lower and upper expectations; and studied some issues related to learning. Another interest is probabilistic satisfiability where propositional sentences are mixed with probabilities, thus specifying sets of probability measures.

Robotics, Computer Vision, etc

Apart from these issues, I have worked in a variety of applications, mostly within Artificial Intelligence and a bit within Robotics and Computer Vision; in the past I have been particularly involved with mobile robots and their sensors.

Welcome to

Fabio Cozman's Home Page!