100 likes | 274 Views
CS 4701 – Practicum in Artificial Intelligence Pre-proposal Presentation. TEAM SKYNET: Brian Nader Stephen Stinson Rei Suzuki. Write a Chinese checkers AI Improve our understanding of… Heuristic Search Machine Learning Other AI features Motivation: TO TERMINATE ALL ENEMIES!.
E N D
CS 4701 – Practicum in Artificial IntelligencePre-proposal Presentation TEAM SKYNET: Brian Nader Stephen Stinson Rei Suzuki
Write a Chinese checkers AI • Improve our understanding of… • Heuristic Search • Machine Learning • Other AI features • Motivation: TO TERMINATE ALL ENEMIES! Problem Statement and Motivation
Server AI Input • 1st input: initial board state • Opponent piece movements • Illegal move errors • Game end condition • AI Output Server • Movement requests Input/Output
Main Reading: Artificial Intelligence: A Modern Approach, Russell and Norvig, Prentice-Hall, Inc. 3rd Ed. • Secondary Sources include, but are not limited to: • The Internet Background Reading
Create a modular AI that makes use of: • Heuristic search functions • Multiple heuristics and heuristic combinations • Hard-coded strategies (opening moves) • Unique strategies General Approach
Heuristic function class • Game state saved at each node • Search algorithm class • Neutral peg placement function • Main([args]) class System Architecture && Work Plan
Chinese checkers statistics available online/reference material • Nightly testing results Data Sources
AI vs. AI games • If simulator is provided • Nightly test results • Wins/Losses • Any other information provided Evaluation Plans
Week 1 (9/11—9/17): Research, proposal completion • Week 2: Design core structure of code base / begin coding • Week 3: Finish coding core structure/ finalize decisions on heuristic and algorithm choices • Week 4: Have prototype AI fully functional (able to move pieces)/ continue developing algorithms • Code Review : Tuesday October 4th • Week 5: Begin testing/optimizing heuristics and searches / searches should be finished • Week 6: Development of unique strategies and ensuring our AI is competitive • Week 7: Testing and finalization of unique strategies • Week 8: Finalize project • Code Review: Tuesday November 1st • Weeks 9-12: Writing final reports and tweaking AI • Final Project Due: November 29th (Judgment Day) Schedule
Neutral Pegs • Does the peg stay there for the remainder of the game? • How many? How will the AI know? • Cross-positions? • Simulator? • Memory constraints • Query the server? • Nightly Testing results • How much info will be provided from the tests? • Board Shape • Holes? • Will the board change from testing to competition? Questions for you!