Ammar designed and trained an agent to play 2048 exploring multiple neural network architectures and hyperparameters. His poster will describe his process of learning and implementing the neural networks in python and analysis of the performance of various reward functions.
This project will only appear in the online showcase.