/reinforcement-learning-F22

Repository that contains lab work for RL-F22 course work in Innopolis University

Primary LanguageJupyter Notebook

reinforcement-learning-F22

Repository that contains assignments for RL-F22 course work in Innopolis University

An Agent based on DQN that moves several cargos across the grid world into a common desirable r ectangle area. There are several cargos, each cargo can be moved either horizontally or vertically by one cell up or down. The size of the overall world may vary, as well as the placement of the cargo and desirable area. The game ends when all the cargos are in the desirable area and do not overlap.

Implemention of building blocks for VGG, GoogleNet, ResNet, ResNeXt and SENet.

Implementation of multi armed bandit and exploring different policies, greedy, epsilon-greedy, softmax and softmax with temprature.

Implementation of SARSA and SARSA(λ) learning algorithms on a windy gridworld environment.