Description:

Project description

System Overview

Universal game AI system. Game developers need to provide functions like those provided by Open-AI game, so they can train their game AI.

Use Instructions

1.For games that use images as state

1. use “ init_UDQN(game, inputImageSize, choose_optimizers, learning_rate) “ to create your network object.

2. the object has the function “run(game, inputImageSize, total_steps, total_reward_list, num, step_num)” to train.

3. “ plot_cost() “ shows the gradient change graph

2.For games that take parameters as state

1. use “ Population(population_num) “ to create your Population object.

2. the object has the function “ initPopulation(self, net_in, net_h1, net_h2, net_out)” to create your network.

3. the object has the function “ runGame(game)” to train.

License

For personal and professional use. You cannot resell or redistribute these repositories in their original state.

Share

• Share on Facebook
• Share on Twitter