Neural_Networks_Project2
Unsupervised and reinforcement learning in neural networks EPFL CS-434
Joachim Muth and Sebastien Speierer January 3, 2016
Reinforcement learning using a rate-based neuron model
For this project, we’re given a code implementing a reinforcement learning paradigm of a rat placed in an arena learning the position of a cheese (its goal). The arena is a discrete grid and the rat is moving step by step. Our goal is to implement a reinforcement learning algorithm using a rate-based neuron model on the same setup such that the position and movement of the rat is not dependent of the grid anymore. In other words, the arena is now a continuous world and the cheese will be represented as a circle area in the arena.