Learning Delayed Influence of Dynamical Systems From Interpretation Transition (PDF)
by Tony Ribeiro, Morgan Magnin and Katsumi Inoue
In some biological and physical phenomena, effects of actions or events appear at some later time points.Such delayed influence can play a major role in various biological systems of crucial importance. Learning such dynamics is the purpose of our work. We propose an extension of the learning from interpretation transition approach that has been proposed previously. This method considers as input a set of state transitions and builds a normal logic program that realizes the given transition relations. The novelty of this work is that the new algorithm we propose is able to consider $k$-step transitions, while the previous one dealt only with 1-step transitions. Hence, we are now able to capture delayed influence with an inductive logic programming methodology.