Research: Symbolic Reinforcement Learning
Paper published in HAL-Inria (Inria’s Research or Technical Reports) during a semester project in my last MSc year.
Abstract
Modeling complex problems is a real issue in Reinforcement learning. The goal of this projet is to use a RL model, such as Q-learning, on symbolic data, where the knowledge of the problem would be stored, in the form of qualitative or numerical data (in JSON for example). Instead of only working with numerical data, we could link the data with a certain ’edition distance’ which permits to evaluate the proximity of 2 elements. This idea has been shared in an international communication and we are trying to explore it with members of the ‘MNEMOSYME’ team at Inria. It is also important to note that those mecanisms are used to model in an operational manner the executive functions of the cortex region in our brain.