La programación lógica inductiva para el aprendizaje automatizado de conceptos
Archivos
Fecha
2011-07-01
Autores
Chaviano Arteaga, Humberto
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad Central “Marta Abreu” de Las Villas
Resumen
En este trabajo se realiza un estudio de otra área del Aprendizaje Automatizado (Machine Learning)
poco utilizada en el contexto de la Ciencia de la Computación en Cuba: la Programación Lógica Inductiva (ILP de sus siglas en inglés). La Programación Lógica Inductiva se inscribe dentro de la disciplina de la Inteligencia Artificial, en el campo del Aprendizaje Automatizado, específicamente en el Aprendizaje Inductivo de Conceptos. Constituye una intersección entre el Aprendizaje Automatizado y la Programación Lógica de primer orden, manteniendo el mismo fin que el primero y aprovechando las potencialidades de la última: a partir del suministro de una serie de ejemplos negativos y positivos se pretende construir una hipótesis como en Aprendizaje Inductivo. La idea básica es cubrir los ejemplos positivos y no los negativos (o solo un número pequeño de estos), utilizando el poder de representación y formalismo de la Lógica de primer orden para poder inducir conceptos representados por programas lógicos. Se hace una descripción de los aspectos teóricos más trascendentales de la Programación Lógica Inductiva: origen, definición, problema que resuelve, clasificación de sistemas, ventajas y desventajas, así como el estudio teórico y práctico de algunos sistemas de ILP más comúnmente utilizados en la comunidad científica. Se incluyen también algunos ejemplos con el fin de que el lector comprenda mejor y de una manera más amena la teoría que se presenta.
In this work is carried out a study of another area of the Machine Learning (ML) little used in the context of the Computer Science in Cuba: the Inductive Logic Programming (ILP of their initials in English). The Inductive Logic Programming registers inside the discipline of the Artificial Intelligence, in the field of the Machine Learning, specifically in the Inductive Learning of Concepts. It constitutes an intersection between the Machine Learning and the Logic Programming of first order, maintaining the same purpose that the first one and taking advantage of the potentialities of the last one: starting from the supply of a series of negative and positive examples it is sought to build a hypothesis like in Inductive Learning. The basic idea is to cover the positive examples and not the negatives (or alone a small number of these), using the representation power and formalism of the Logic of first order to be able to induce concepts represented by logic programs. A description of the most momentous theoretical aspects in the Inductive Logic Programming is made: origin, definition, problem that solves, classification of systems, advantages and disadvantages, as well as the theoretical and practical study of some systems of ILP more commonly used in the scientific community. They are also included some examples with the purpose of that the reader understands better and in a more interesting way the theory that is presented.
In this work is carried out a study of another area of the Machine Learning (ML) little used in the context of the Computer Science in Cuba: the Inductive Logic Programming (ILP of their initials in English). The Inductive Logic Programming registers inside the discipline of the Artificial Intelligence, in the field of the Machine Learning, specifically in the Inductive Learning of Concepts. It constitutes an intersection between the Machine Learning and the Logic Programming of first order, maintaining the same purpose that the first one and taking advantage of the potentialities of the last one: starting from the supply of a series of negative and positive examples it is sought to build a hypothesis like in Inductive Learning. The basic idea is to cover the positive examples and not the negatives (or alone a small number of these), using the representation power and formalism of the Logic of first order to be able to induce concepts represented by logic programs. A description of the most momentous theoretical aspects in the Inductive Logic Programming is made: origin, definition, problem that solves, classification of systems, advantages and disadvantages, as well as the theoretical and practical study of some systems of ILP more commonly used in the scientific community. They are also included some examples with the purpose of that the reader understands better and in a more interesting way the theory that is presented.
Descripción
Palabras clave
Programación Lógica Inductiva, Aprendizaje Automatizado, Aprendizaje de Conceptos, Inteligencia Artificial