Skip to content

Latest commit

 

History

History
31 lines (21 loc) · 1.06 KB

File metadata and controls

31 lines (21 loc) · 1.06 KB

Reinforcement Learning Simulation

Demo

Reinforcement

About

This Simulation Implements the Concept of Reinforcement learning, where a Car "learns" to avoid obstacles by Trial and Error, and gets better with Time at achieving its End goal.

What is Reinforcement Learning

  • In Supervised Learning, there is an "answer" for "data". You can decide upon some method and build a model to be able to predict upon unknown values.
  • In Reinforcement learning, There is no way to decide upon a fixed model.
  • All you know is - You have to reach a goal.
  • How ? Trial and Error.
  • Consistst of an agent, and no other useful data to help it achieve its goal.
  • It learns from "experience" - collecting training examples through Trial and error as it attempts to reach the goal.
  • For every right step, the agent is rewarded
  • For every mistake, agent is punished.
Best example : Self Driving Car

Main script

Built With

  • Basic Javascript,HTML,CSS