shangeth / Lunar-Lander-Deep-RL

Deep Q network, DDQN, Dueling DQN for Lunar Lander Environment.

Lunar Lander DRL

Environment

Reinforcement Learning Approaches

• Discrete space - Traditional RL methods like MC, TD

• Continuous Space

  • Discritize the space

    • Grid
    • Tile Coding
    • Coarse Coding

  • Continuous Function Approximation

    • Linear Approximation
    • kernel Approximation
    • Non-Linear Approximation

Non-Linear Approximation

Deep Q Network - Paper

Double Deep Q Network - Paper

Prioritized Experience Replay - Paper

Dueling Deep Q Network - Paper