ChessAI

Repository containing Chess Engine and ML Model

TODO

Wasif

• compare separated vs non-separated network
• get gpu acceleration
• tweak rewards (inputs) to see best one-prioritize defense
• custom loss function (outputs) to capture more chess strategy
  • GradientTapes w/ custom training loop

• MILESTONE 1.5 base smart AI done

• Set up RNN/cnnRNN,CNN/Greedy/Greedy+exploit
• Move w.r.t. various reward functions
• Set up qlearning (links game to models>)
• Train/Test
• IReinfLearn to learn reward

MILESTONE 2: Smart AI done

• Submit to Lichess

Ali

• unit tests for board and game
• Write actual integration tests (one for why moves needs to be updated after each move)
  • i think it only needs tobe updated before a king moves
• write 2 system tests
• implement scoreboard + visualization
  • db contains players=(ID,name) and scores=(ID,ID, score)
• type documentation + doc strings
• optimizations
  • change coordinate system to 2-digit int
  • remove if x.is_white: ... else: ... by inverting board at start/end of code
• write AI - X-d,All-b=2s thinking time

MILESTONE 1: Done backend

• Start frontend
• Write Controller to interact Model and View
• Write Buttons for View

MILESTONE 2: Frontend done

Zuraze

• Lose to AI

Project Complete