• 'forest_simulation' is the previous name. The formal name of this project should be 'Hunger Game 2.0'
• In Hunger Game 2.0, we aims to find the optimal policy for an animal to survive as long as possible in a forest.
• The animal gains reward if it finds food, or is punished if it encounters carnivores.
• Implemented 3 different algorithms - traditional Q-learning, Deep Q Network
• with Multilayer Perceptron, Deep Q Network with Convolutional Neuron Network.
• Refined the traditional Q-Learning, with which the animal can survive almost infinite rounds.