Évaluation de l'apprentissage incrémental par analogie (Incremental Learning From Scratch Using Analogical Reasoning )[In French].

Incremental Learning From Scratch Using Analogical Reasoning This paper explores the application of analogical reasoning on incremental machine learning. The learning task is about transferring natural language requests to formal language commands. This work explores two questions for applying analogy in incremental learning situation : How does the analogical reasoning behave in incremental learning situation ? How do the conditions on the learning sequence influence the performance ? To address this issue, an experimental setup is proposed in which two different users are simulated with various learning sequences. We point out that the order of the incoming learning examples has a notable influence only under particular conditions. The experiments also show that incremental learning and analogical reasoning are complementary for a small number of learning examples. MOTS-CLÉS : apprentissage incrémental, raisonnement analogique, transfert de langage.