This preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: 1 Course 45: Informed search (Heuristics) algorithms Outline ¡ Bestfirst search ¢ Greedy bestfirst search ¢ A * search ¡ Local search algorithms ¢ Hillclimbing search ¢ Simulated annealing search ¢ Local beam search ¢ Genetic algorithms ¡ Basic Optimization Problem and Solution ¢ Linear Programming Search Algorithms ¡ Blind search – BFS, DFS, uniform cost ¢ no notion concept of the “right direction” ¢ can only recognize goal once it’s achieved ¢ A search strategy is defined by picking the order of node expansion ¡ Heuristic search – we have rough idea of how good various states are, and use this knowledge to guide our search 2 Bestfirst search ¡ General approach of informed search: ¢ Bestfirst search: node is selected for expansion based on an evaluation function f(n) ¡ Idea: evaluation function measures distance to the goal. ¢ Choose node which appears best ¡ Implementation: ¢ fringe is queue sorted in decreasing order of desirability. ¢ Special cases: greedy search, A* search A heuristic function ¡ [dictionary] “A rule of thumb, simplification, or educated guess that reduces or limits the search for solutions in domains that are difficult and poorly understood.” ¢ h(n) = estimated cost of the cheapest path from node n to goal node . ¢ If n is goal then h(n)=0 How to design heuristics? Later.. 3 ¡ An A algorithm is a bestfirst search algorithm that aims at minimising the total cost along a path from start to goal. ¢ f(n) = g(n) + h(n) estimate of total estimate of total cost along path cost along path through n through n estimate of estimate of cost to reach cost to reach goal from n goal from n actual cost to actual cost to reach n reach n A heuristic function Romania with step costs in km ¡ h SLD =straightline distance heuristic. ¡ h SLD can NOT be computed from the problem description itself ¡ In this example f(n)=h(n) ¢ Expand node that is closest to goal ¢ greedy bestfirst search 4 Romania with step costs in km Greedy bestfirst search ¡ Evaluation function f(n) = h(n) ( h euristic) ¡ = estimate of cost from n to goal . ¡ e.g., h SLD (n) = straightline distance from n to Bucharest. ¡ Greedy bestfirst search expands the node that appears to be closest to goal. 5 Greedy bestfirst search example Greedy bestfirst search example 6 Greedy bestfirst search example Greedy bestfirst search example 7 Properties of greedy bestfirst search ¡ Complete? No – can get stuck in loops, e.g., Iasi Æ Neamt Æ Iasi Æ Neamt Æ ¡ Time? O(b m ) , but a good heuristic can give dramatic improvement ¡ Space? O(b m )  keeps all nodes in memory ¡ Optimal? No A * search ¡ Idea: avoid expanding paths that are already expensive....
View
Full Document
 Fall '09
 ahmetyazıcı
 Search algorithms, Search algorithm, A* search algorithm, Bestfirst search

Click to edit the document details