36-人工智慧搜寻演算法与解题实作.ppt

國立台北大學 資訊工程學系 人工智慧搜尋演算法與解題實作 Jen-Chun Chang 張仁俊 (modified from 陳榮傑) Background Data structures Stacks, queues, hash tables, heaps … Algorithms Recursive algorithms Dynamic programming 3 classes of AI search problems Problem Space Model 八位數碼問題(8 puzzle) How to avoid duplicate node in the search tree? Use hash! Search strategies Brute-Force Search Breadth-First Search (BFS) Uniform-Cost Search Depth-First Search (DFS) Depth-First Iterative-Deepening Search (DFID Search) Bidirectional Search Order of node generation forbreadth-first search Order of node generation fordepth-first search Order of node generation for DFID search Search strategies Heuristic Search Pure Heuristic Search A* Algorithm Iterative-Deepening-A* Depth-First Branch-and-Bound Heuristic Path Algorithm Best-First Search In each cycle of a best-first search, the node n that is best according to some cost function f(n) is chosen for expansion. Let g(n) be the cost of the path from the initial state to node n, and h(n) be the heuristic estimate of the cost of a path from node n to a goal. Best-First Search Breadth-first search, uniform-cost search, and pure heuristic search are all special cases of best-first search. Breadth-first search:f(n)=the depth of node n Uniform-cost search: f(n)= g(n) Pure heuristic search: f(n)= h(n) A* algorithm: f(n)= g(n)+h(n) Heuristic path algorithm: f(n)= (1-w)*g(n)+w*h(n) 量杯問題 三量杯容量 80, 50, 30 80杯滿, 50, 30杯空 如何得使三量杯得水 40, 40, 0? 量杯問題(BFS) How to implement BFS? Use queue! 八位數碼問題(DFS) How to implement DFS? Use stack! Node expansion: Generating all the children of a given state. Node generation: Creating the data structure that corresponds to a problem state. DFS always generates next a child of the deepest unexpanded node. 旅行者推銷員問題(uniform-cost search) How to implement Best First Search? Use priority queue(heap)! 反溯法Backtracking 八后問題The 8-queen problem C