AI Learns How to Play Pong - Reinforcement Learning

preview_player
Показать описание
This video shows how deep reinforcement learning agent learns to play against a hard-coded bot. DeepQ architecture is used for the AI agent. At first, AI had two available actions: Up, Down. Due to the unstable behavior, the final AI has three actions: Up, Do Nothing, Down.

Don't forget to subscribe and click on the bell :)
  1. ML Guy
  2. ML
  3. Reinforcement Learning
  4. Deep Learning
  5. AI
  6. MLGuy
Рекомендации по теме
AI Learns to 0:21:30

AI Learns to Speedrun Minecraft

AI Learns to 0:10:29

AI Learns to Play Tag (and breaks the game)

AI Learns To 0:13:57

AI Learns To Play Golf

AI Learns To 0:10:31

AI Learns To Play CS:GO By Watching Humans Play!

OpenAI Plays Hide 0:06:02

OpenAI Plays Hide and Seek…and Breaks The Game! 🤖

A.I Learns to 0:11:32

A.I Learns to Play TOWER DEFENSE

AI Learns to 0:08:07

AI Learns to Speedrun Mario

AI Learns Insane 0:11:30

AI Learns Insane Monopoly Strategies

Persuading with Generative 0:00:38

Persuading with Generative AI: Succesfully Overcoming Objections

AI Learns to 0:11:51

AI Learns to play Geometry Dash || FINAL

AI Learns to 0:29:25

AI Learns to Play Wii Sports

AI Learns To 0:12:56

AI Learns To Play Lethal Company

Training AI to 0:33:53

Training AI to Play Pokemon with Reinforcement Learning

AI Learns to 0:09:54

AI Learns to DESTROY old CPUs | Mario Kart Wii

AI Learns to 0:06:54

AI Learns to play the Worlds Hardest Game

AI Learns Table 0:11:20

AI Learns Table Tennis

Can AI play 0:10:06

Can AI play a track it's never seen before? | Mario Kart Wii

A.I Learns to 0:17:14

A.I Learns to Play DODGE BALL

I Made 1.000 0:11:50

I Made 1.000 A.I Warriors FIGHT... (Deep Reinforcement Learning)

AI Invents New 0:11:33

AI Invents New Bowling Techniques

Evolving Genetic Neural 0:09:59

Evolving Genetic Neural Network Optimizes Poly Bridge Problems

Multi-Agent Hide and 0:02:58

Multi-Agent Hide and Seek

AI Olympics (multi-agent 0:11:13

AI Olympics (multi-agent reinforcement learning)

Training an unbeatable 0:20:41

Training an unbeatable AI in Trackmania

Комментарии
Автор

hey I am having issues with it, I have also made my own, do you have discord? bcz my model is not learning anything, and I am getting frustated now because it's been since 3 weeks not I am training!

AbhayShinde-fzwd
Автор

Yo how long did it take to train? Overnight I was able to train a somewhat good pong ai going against a hard coded one that follows the ball, but I wanted to do more so now I'm trying out self play and I think I know why you decided not to do that, they're really fucking stupid XD. Good luck in your endevors

praxition
Автор

Great!! I suppose you used raw data from the game memory, not visual. Because my agent has similar behavior. Do you remember what type of data you passed to it (ball_x, paddle_y)? Or maybe the model and if it was one with how many layers?

anseg
Автор

Hey can you share the code you wrote for this ? thank you

vent_srikar