1 / 21

Informed (Heuristic) Search Strategies

Informed (Heuristic) Search Strategies. By: Zac Lippard CS 430. What is a Heuristic Search?. It is an informed type of strategy Uses information for problem-solving Can find solutions faster than uninformed strategies. The 3 Basic Heuristic Search Strategies. Greedy Best-First Search

tavon
Download Presentation

Informed (Heuristic) Search Strategies

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Informed (Heuristic) Search Strategies By: Zac Lippard CS 430

  2. What is a Heuristic Search? • It is an informed type of strategy • Uses information for problem-solving • Can find solutions faster than uninformed strategies.

  3. The 3 Basic Heuristic Search Strategies • Greedy Best-First Search • A* Search • Memory-Bound Heuristic Search

  4. Understanding the Search Method • Takes in information in order to find the quickest solution. • Solution should also be efficient. • Not every method will be efficient depending on the algorithms used.

  5. Functions Behind the Searches • Evaluation Function • Heuristic Function

  6. Evaluation Function • The search algorithm that chooses a node best suited for expansion. • Most functions search for nodes with smallest value.

  7. Heuristic Function • The algorithm used to find the cheapest path to the goal node. • Most common form of search algorithm. • Easier to use • More information readily available

  8. Greedy Best-First Search • Expands the node that is closest to the goal. • Evaluates nodes by the function:f(n) = h(n) • Example is a straight-line distance heuristic • Uses straight lines with the shortest distance to reach the goal.

  9. Problems with this Search • The main problem with this search is even though the algorithm picks shortest distances, it doesn’t necessarily find the shortest distance to the goal.

  10. Example of a straight-line distance

  11. Greedy Best-First Search Expansion

  12. A* Search • Most widely known form of the best-first search • The evaluation function is: f(n) = g(n) + h(n) • g(n) is the cost to reach the node. • The cheapest solution is found by g(n) + h(n)

  13. A* Search Cont. • Admissible Heuristic = Optimistic search • It never overestimates; estimates the cost less than actuality. • Contours can be used to find the general range of possible nodes that can be expanded given a certain f-cost.

  14. A* Optimality

  15. A* Search Example

  16. Memory-Bound Heuristic Search • Used to reduce the amount of memory used in the search function. • The IDA*, or iterative-deepening A* algorithm, has a cutoff value of the smallest f-cost from the previous iteration.

  17. Recursive best-first search (RBFS) • A better version of IDA*. It regenerates nodes from the same level in order to find the smallest f-cost. • The recursive nature from regenerating the same node over and over is one of the downfalls.

  18. Memory-bounded A* (MA*) • This search uses an optimal amount of memory. • A simpler form of MA* is Simplified MA* or SMA*.

  19. Example from book • Enhances the RBFS function by removing the highest f-cost node from memory in order to make room for new expanded nodes. • What if there are multiple highest f-cost with the same value? • SMA resolves this by expanding the newest best leaf node and deleting the oldest one.

  20. Better Searches • How can we have better searches? • Through meta-level state space. • Meta-level structures allow the computer to capture the internal state of an object-level state (in the books example this would be Romania). • Errors like unnecessary node expansion are interpreted by a meta-level learning algorithm.

  21. Conclusion • Types of known search strategies: • Greedy Best-First Search • A* Search • MA* and SMA* • Meta-level Searches • Goal is to always minimize the total cost of the problem.

More Related