- Compared various agents like Sarsa(0),Expected Sarsa and Q-Learning on the ‘Windy Grid world’ problem.
- Experimented with LSTM-A3C model to develop an AI for playing Breakout (arcade game) and trying to beat it
- Developed an AI, a Deep Convolutional Q-Learning model combined to Eligibility Traces for playing ’Doom’ which is a first person shooting game and trying to beat the high score