The guest blogger for today is our colleague  Moises Goldszmidt from MSR-SVC  who was Judea Pearl ‘s student from 88 to 92  (a couple of related posts can be found here and here):

———————————————————————-

In celebration of Judea Pearl winning the 2011 A.M. Turing Award I would like to provide a personal view and perspective on a couple of Judea’s key insights.

Embracing uncertainty was becoming a necessity for anybody interested in embedding intelligent systems in realistic environments during the 80’s.  The task was full of challenges and obstacles.  For a while researchers were questioning the suitability and power of probability theory giving rise to a set of alternative calculi and theories.  For Judea probability was just fine. His fundamental insight was that a large obstacle resided in the lack of expressive means for articulating known or learned structure and relationships between probabilistic entities.  Judea (and students and collaborators) focused then on providing those capabilities using directed acyclic graphs to represent “influences” between random variables.  The legacy of this research program includes not only the introduction of Bayesian networks, but a sound axiomatization of conditional independence and an efficient set of algorithms (and heuristics) for marginalization  and conditioning; both fundamental operations in manipulating probability statements.

Embracing causality has been controversial in philosophy and the sciences (I am including statistics, economics, and the social sciences in here) and proven difficult to formalize.  Yet, important questions affecting our society and well-being beg the ability to reason with causal relationships and make causal inferences: What are the consequences of a policy change?  What is the root-cause of a failure? What would have happened had I administered half the dose of a drug?  Judea’s insight here was to notice that at the core, these questions assume an intervention of some sort.  Thus, he proceeded to investigate the mathematics behind a calculus for intervention, again relying on an innate computer science “data structure”: graphs.  By this route he was able to make inroads into causality providing part of the mathematics needed to address these problems from a computational perspective.

Embracing curiosity is a trademark of researchers we admire and respect.  Judea is no exception.  He possesses an insatiable passion for understanding and making things better.  The contributions in his body of work range from the conceptual and philosophical to the extremely practical. May that kind of passion infect and guide our own explorations and provide the seed to fruitful and exciting results and discoveries.