Basic AI lecture in 2021-2 semester

• Learning a Basic AI course in 2021-2 semester
• Reference - Russell, S. and Norvig, P. (2021). Artificial Intelligence: A Modern Approach (4th edition). Pearson.
• using python

pj1. maze

stage1 - bfs vs astar

size	bfs (path 길이 / search states)	astar (path 길이 / search states)
small	9 / 16	9 / 14
medium	69 / 274	69 / 221
big	211 / 647	211 / 548

stage2 - heuristic function using manhatten dist vs using path length

metric	mst using manhatten dist as edge cost -- (1)	mst using path length as edge cost -- (2)
state 개수	1901 (stage2 - big.txt 기준)	49 (stage2 - big.txt 기준)
수행시간	0.049 sec (stage2 - big.txt 기준)	1.126 sec (stage2 - big.txt 기준)
stage3 - big size	해결 못함	해결 함

pj2. pacman

정리

	minimax	alphabeta
iteration 300, smallmap, depth=2	Tot = 48.28 Avg = 0.16	Tot = 40.82 Avg = 0.13
iteration 300, mediummap, depth=2	Tot = 121.16 Avg = 0.403	Tot = 99.17 Avg = 0.33
iteration 1000, minimaxmap, depth=4	Tot = 75.23 Avg = 0.07	Tot = 41.55 Avg = 0.04

=> alphabeta agent가 minimax agent보다 효율적임을 알 수 있다. depth를 깊게 실험해보면, alphabeta agent의 pruning 효과가 더욱 극명하다.

pj3. logical agent

pj4. MLP vs CNN