A Deep Reinforcement Learning problem of a Self-driving car was opened as a competiton.

Registered people had to make changes in the given Network so that it can achieve max avergae speed in MPH.

In version 1.0 I reached to 10th position by getting help from different sources and using my basic Deep RL knowledge so far.

The best part was looking at Convnetjs which is made for practising Deep Learning and RL algorithms by the power of javascript inside browser without Linux or GPU support.

For the code here is the repository.