Search

We discuss a procedure which extracts statistical and entropic information from data in order to discover Boolean rules underlying them. We work within a granular computing framework where logical implications between statistics on the observed sample and properties on the whole data population are stressed in terms of both probabilistic and possibilistic measures of the inferred rules. With the main constraint that the class of rules is not known in advance, we split the building of the hypotheses on them in various levels of increasing description complexity, balancing the feasibility of the learning procedure with the understandability and reliability of the formulas that are discovered. We appreciate the entire learning system in terms of truth tables, formula lengths, and computational resources through a set of case studies. Index Terms--Algorithmic inference, Boolean formula simplification, computational learning, fuzzy sets, granular computing, mutual information, probably approximately correct (PAC) meditation, rough sets, rule learning, sentry points.

With the aim of getting understandable symbolic rules to explain a given phenomenon, we split the task of learning these rules from sensory data in two phases: a multilayer perceptron maps features into propositional variables and a set of subsequent layers operated by a PAC-like algorithm learns Boolean expressions on these variables. The special features of this procedure are that: i) the neural network is trained to produce a Boolean output having the principal task of discriminating between classes of inputs; ii) the symbolic part is directed to compute rules within a family that is not known a priori; iii) the welding point between the two learning systems is represented by a feedback based on a suitability evaluation of the computed rules. The procedure we propose is based on a computational learning paradigm set up recently in some papers in the fields of theoretical computer science, artificial intelligence and cognitive systems. The present article focuses on information management aspects of the procedure. We deal with the lack of prior information about the rules through learning strategies that affect both the meaning of the variables and the description length of the rules into which they combine. The paper uses the task of learning to formally discriminate among several emotional states as bath a working example and a test bench for a comparison with previous symbolic and subsymbolic methods in the field. Index Terms--Edge pulling functions, fuzzy relaxations, hybrid systems, learning fitness, Ockham razor, PAC-learning, symbolic feedbacks, understandable learning.