UIB-AILS: A Model for Autonomous Inductive Learning Systems

Systems that learn autonomously constitute a relevant exponent with-in the category of intelligent systems.Such systems are characterized by the ability to update their decision/action hypothesis over time without any external intervention.Such capacity for updating is based on some main considerations, such as: the experiences accumulated by the system, the acquisition of new knowledge from the environment, the general structural characteristics of the hypothesis and the criterion specifying the particular properties that the hypoth-esis must satisfy.The design of an autonomous inductive learning system -AILS -is a complex task.This work presents a general model for the design of AILSs, whose components can be customized according to the nature of the problem in hand, so that the model is suitable for addressing the design of a variety of AILS with different peculiarities.Such a model adopts a cyclic evolutionary configuration that includes, among others, components to deal with data imprecision, tech-niques to handle the vagueness of decisions/actions, and methods to process in a unified way knowledge coming from different levels of abstraction, such as raw data and logic expressions.