The framework employs a physics simulation in aprobabilistic way to infer about moves performed by an agent in a setting governed by Newtonian laws of motion . We complement the model with a model-free approach to aid the samplingprocedures in becoming more efficient through learning from experience during game playing . We discuss the performance of themodel on the game of Flappy Bird. We present an approach where combining model-based approaches (aconvolutional neural network in our model) is able to achieve what neither could . alone . This way the model outperforms an all model free or all model-driven approach.This way the models outperforms . an all . model free approach to outperform an allmodel-free or all . approach .

Author(s) : Fahad Alhasoun, Sarah Alnegheimish, Joshua Tenenbaum

Links : PDF - Abstract

Code :

https://github.com/oktantod/RoboND-DeepLearning-Project


Coursera

Keywords : model - approach - free - game - physics -

Leave a Reply

Your email address will not be published. Required fields are marked *