/project
In this project, we applied the World Models Deep Learning Framework to an Arcade Game Space Invaders.
World Models is a Deep Learning framework which trains a network much like how a human learns about the world, by focussing on the idea of cultivating a model's internal mental model of their environment such that the model can train using it's own imagination.
My specific section was training Mixed Density Recurrent Neural Network to predict the next game state and action.