人工智能：一种现代方法ch03-2 solving problems by searching - 2016.pdfVIP

有信息搜索（启发式搜索） 1 启发式函数：如何更快地接近目标？什么是启发式函 数？什么是启发式搜索？ 2 贪婪最佳优先搜索，A*搜索 3 可采纳性、一致性、准确性、松弛问题：什么是可采 纳的启发函数？什么是一致的启发函数? 启发式函数的 准确性。如何设计可采纳的启发函数？ 1 review 2 Blind Strategies  Breadth-first ：depth(n)  Uniform-cost: g(n)  Depth-first: -depth(n)  Depth-limited: -depth(n), depth(n)=limit  Iterative deepening  Speeding up techniques  Avoiding repetitive states  Bi-directional search  Blind strategies do not know which node in frontier is closer to the goal. 6 启发式函数 7 How do we know which node in frontier is closer to the goal? 8 An Idea: estimating the distance to the goal  It would be better if the agent knew whether or not the city it is travelling to gets it closer to Bucharest  Of course, the agent doesn’t know the exact distance or path to Bucharest (it wouldn’t need to search otherwise!)  The agent must estimate the distance to Bucharest 9 Informed (Heuristic) search {3.5}  More generally  We want the search algorithm to be able to estimate the path cost from the current node to the goal  This estimate is called a heuristic(启发式) function  Can’t be done based on problem formulation  Need to add additional information  Informed search 10 Heuristic Function {3.5}  Heuristic function h(n)  h(n): estimated cost from node n to goal  h(n