Search Summary Search Summary Some material from: D Lin, J You, JC - - PowerPoint PPT Presentation

search summary search summary
SMART_READER_LITE
LIVE PREVIEW

Search Summary Search Summary Some material from: D Lin, J You, JC - - PowerPoint PPT Presentation

RN, Chapter 25 Search Summary Search Summary Some material from: D Lin, J You, JC Latombe 1 Search Summary # 1 Problem Solving as Search Blind Search Techniques Breadth-first (uniform cost) Depth-first Iterative


slide-1
SLIDE 1

1

Search Summary Search Summary

RN, Chapter 2—5

Some material from: D Lin, J You, JC Latombe

slide-2
SLIDE 2

2

Search Summary # 1

Problem Solving as Search Blind Search Techniques

Breadth-first (uniform cost) Depth-first “Iterative Deepening” Bi-Directional Time/ Space Complexity:

Size of search space: ≈ 1011 nodes

slide-3
SLIDE 3

3

Search Summary # 2

Heuristic Search Techniques

… using “Distance to Goal”

Best-First A* : provably optimal!

Search space ≈ 1025 nodes (IDA*)

Heuristic Functions

slide-4
SLIDE 4

4

Search Summary # 3

Constraint Satisfaction Problems

Intro CSP (Def’n, Types, Examples) Complexity Tricks for “Grow” approach

(arc) consistency + probagation Backward checking (DFS) Forward Checking Variable / Value ordering

Constraint Optimization Problems

slide-5
SLIDE 5

5

Search Summary # 4

Iterative Algorithms

Framework, Examples Hill-climbing / Gradient Descent Problem / Issues GSAT, WalkSat Other approaches

Simulated Annealing, Tabu, Random Restarts,

Genetic Algorithms

⇒ Search space ≈ 10100 to 101000

slide-6
SLIDE 6

6

Search Summary # 5

Adversary Saerch / Game Playing

Minimax

≈ 1010 nodes, 6-7 ply in chess

Alpha-beta Pruning

≈ 1020 nodes, 14 ply in chess

provably “optimal”

slide-7
SLIDE 7

7

Other Topics

wrt Search

Iterative BROADENING Memory Bounded Search – SMA* Beam Search Island Hopping – abstraction …

wrt CSPs

Backjump Dynamic Orderings Special cases (eg, when arc-consistency is

sufficient)

slide-8
SLIDE 8

8

Search and AI

Q: Why such a central role? A: As many AI tasks are

ill-specified and/or intractable,

Search is ONLY approach

Many applications of search:

Learning, Reasoning, Planning, Design, GamesPlaying, NLU, Vision, …

Good news:

Tremendous recent progress 1030 feasible; often to 101000

QUALITATIVE DIFFERENCE from only a few

years ago!!