ANN Browser-based Game

This browser-based game, that mocks Chrome Dinosaur Game, runs client-side artificial neural networks to train the game character and then play autonomously.

Find the code here:
https://github.com/arjunsk/phaser_ai_game

Transcript

1. Jumper - AI Game
2. >Libraries used ▸ PhaserJS game framework ▸ Synaptic neural network library in javascript
3. >BACKGROUND ▸ Perceptron having 1 input, 2 hidden & 1 output layer ▸ Supervised Learning using manual input
4. >INPUTS ▸ velocity of the ball ▸ displacement between ball & player
5. >OUTPUTS ▸ ON FLOOR ▸ ON AIR
6. INPUT OUTPUT HIDDEN LAYER 2 HIDDEN LAYER 1 DISPLACEMENT VELOCITY ON AIR ON FLOOR
7. >SCENARIOS ▸ False user input ▸ No user input ▸ Correct user input
8. >DEFUZIFYING TECHNIQUE Maximum membership principle
9. >INSPIRED FROM ▸ Chrome’s T-REX
10. >CODE ▸ Initializing ANN
11. >CODE ▸ Training ANN
12. >CODE ▸ Activating ANN and getting the output
13. >EXIT 0 THANK YOU :)