iOthello

Tasks

Further Studies
Policy Network
Policy network could be devised to have a better response time.
- day
Tree Traversal
Monte Carlo Tree Search and other tree traversal techniques can be attempted.
- day
Reinforcement Learning
Learning players move and learning by its own mistake should be an attemp worth trying.
- day
Enhancements
Modularity
The components should be completely separable and lie on differnt thread, communcating using signal.
- day
User Control
User should be able to undo and even play at any move ahead. He or she should also be able to print out move history.
- day
Online Runs
Online runs would be extremely beneficial for attracting human testers.
- day
Completed
Background Reading
Finish the reading on "Mastering the game of Go with deep neural networks and tree search" for basic understanding on AlphaGo's underlying principles.
20/09/16
- day
Project Plan
Detailed project plan listing background researched, objectives, methodology and schedule.
02/10/16
- day
Project Website
Simiple project website showing simple information, introduction, acheivements and documentation.
06/10/16
- day
Simple Skeleton
An AI Othello program with GUI to interact with human players. Monte-Carlo Tree Search should be prioritized to mimic AlphaGo and lay the foundation for further development.
09/10/16
- day
Value Network
A value network shoud be developed using play history of either professionals or 200 games of 4 undergraudates.
31/12/16
- day
Interim Report
Objective, background, methodology are still needed, however there should be a new section on what has been accomplished.
22/01/17
- day
Preliminary Implementation
The program should be more or less functional and beyond elementary playing level. It should not commit obvious mistakes.
22/01/17
- day
Finalized Implementation
The program should be fully functional.
31/03/17
- day
Final Report
Literature review, experiments, results and future prospects should be included in this report.
16/04/17
- day

Timeline

Phase 1 Traditional AI Approach in a 8x8 Othello Game.
31/10/16
Phase 2 Deep learning value evaluation network
31/12/16
Phase 3 Extended Research
31/03/17
Phase 4 Final Product
17/04/2017