2019 秋 電腦對局理論 助教

這是個在台大電腦對局理論擔任助教的個人紀錄。除了要向世界表達我擔任助教的開心與成就感之外,大概就是想鼓勵其他人如果有機會當助教的話千萬不要錯過了。
要感謝徐讚昇老師願意給我這次機會當助教,還允許我在同時間修撞課的編譯器設計。也謝謝宗佑不修課還願意旁聽並聽我上台解釋作業耍猴戲。

Continue reading “2019 秋 電腦對局理論 助教”

Einstein Würfelt Nicht Agent

This is variation of Einstein Würfelt Nicht which rules are modified for bigger branching factor and longer game length. Both player starts with 6 game pieces each. The goal is to move the smallest piece possible towards the diagonal of your opponent. This is a 2-agent game to challenge AI agent to do decent pruning and also good evaluation of current board for their Monte Carlo Search. You are also welcome to solve the game with Alpha-Beta Search.

Checkout my repository for the game: 2 agent game – Einstein Würfelt Nicht Agent

Continue reading “Einstein Würfelt Nicht Agent”

Pukoban Solver

Sokoban(推箱子) is a classic single agent game. It has a sufficient of game state and is a classic problem in the field of game AI.
Pokoban is a variation of the game, adding “Pull” to the game. You can pull a box when there is an empty space behind you and a box in front. This additional action gives the game even wider branching factor, making the search harder.
Here is the game interface that you can actually play the game:

Checkout my repository for the game: Single agent game – Pukoban

Continue reading “Pukoban Solver”