January 20th (Wednesday) from 12:30 to 14:00
Seminar Room 2, 15F, NII
Two Enhancements on Abductive Reasoning
Abduction is inference to the explanation, and has been applied
to many problems in AI and has recently become an important
technique for automated scientific discovery. In this talk, I
present two recent work on abductive reasoning.
The first topic [1] is concerned with finding best hypotheses
among many logically possible hypotheses obtained from the
process of hypothesis generation. We propose an abductive
inference architecture combined with an EM algorithm working on
binary decision diagrams (BDDs). An implemented system has
been applied to inference of inhibition in metabolic pathways
in the domain of systems biology.
The second topic [2] addresses discovery of unknown relations
from incomplete network data by abduction. To this end, we
introduce a framework of meta-level abduction, which is powerful
enough to infer missing rules, missing facts, and unknown causes
that involve predicate invention in the form of existentially
quantified hypotheses. This technique has been applied to
discover physical techniques and related integrity constraints
in cognitive modeling of cello playing.
[1] Inoue et al., Proc. 21st IJCAI, pp.810-815, 2009.
http://ijcai.org/papers09/Papers/IJCAI09-139.pdf
[2] Inoue et al., Post-proc. 19th ILP, LNAI, Springer, 2010.
An earlier version can be obtained from:
http://www.cs.kuleuven.be/~dtai/ilp-mlg-srl/dokuwiki/doku.php?id=paper:ilp:15