Deep Q Networks | Q Learning | Reinforcement Learning | Epsilon-Greedy Policy | Python | AI Gym

===== Likes: 21 👍: Dislikes: 0 👎: 100.0% : Updated on 01-21-2023 11:57:17 EST =====
Curious what Q Learning is? Ever wonder how to apply Q learning with Deep Q Networks with an actual example? Well, look no further as I walkthrough step by step on how to apply DQN practices on the ever-popular dinosaur game!

Check out how to setup a custom AI Gym:
https://youtu.be/TY0fyHmCGps

Policy: Epsilon-Greed Policy
Loss: MSE (For reasons specified in video)
Model: DQN (3 Convolution layers, 512 Fully connected network, 2 output nodes)

Github: https://github.com/SpencerPao/ComputerVision/tree/main/Reinforcement_Learning

Check out these resources as well! Lots of inspiration and help came here.
----------------------------
https://pytorch.org/tutorials/intermediate/reinforcement_q_learning.html
https://medium.com/@gtnjuvin/my-journey-into-deep-q-learning-with-keras-and-gym-3e779cc12762
https://github.com/openai/gym/blob/master/gym/envs/classic_control/cartpole.py
https://blog.paperspace.com/dino-run/
https://github.com/moduIo/Deep-Q-network/blob/master/DQN.ipynb
----------------------------

0:00 - RL Definitions & Objectives
0:51 - Q Learning & DQN
1:22 - DQN Process
1:59 - Walkthrough of Environment Class
4:06 - Walkthrough of Agent/Model Class
8:59 - Run Function (Bringing everything together)
12:49 - RL in process

Видео Deep Q Networks | Q Learning | Reinforcement Learning | Epsilon-Greedy Policy | Python | AI Gym канала Spencer Pao