It often happens that, in the revision of one’s work, fresh new ideas swiftly come to mind. After spending time reorganizing this website, it became apparent that—had I had the time—I would attempt to explore new extensions which connect topics that have always interested me, such as the analyses of algorithms, logic and constraint programming, parsing, compiling, and parallelism.
The inspiration for such developments emerged from my view—motivated by bioinformatics—that probability and statistics concepts have often been neglected in most computer science approaches. An exception is the remarkable work of Judea Pearl based on Bayesian networks. Two papers by Russell and Van den Broek [Rus 15, VdB 13] suggest new venues that I hope young researchers will explore.
The basic idea is to incorporate Bayesian inference to logic programming, thus enabling the development of probabilistic versions of constraint programs, program analyzers, parsers and parallel programs.
In retrospect, if I had to restart my career in the sciences, I would concentrate on the fields of mathematics and biology.
[Rus 15] S. Russell, Unifying Logic and Probability, Communications of the ACM, 58 (7), 88-97, 2015.
[VdB 13] G. Van den Broeck, Lifted Inference and Learning in Statistical Relational Models, Doctor in Engineering dissertation, Arenberg Doctoral School, KU (Catholic University) Leuven, Belgium, 2013.