Hill-climbing Search

1
Hill-climbing Search

• Goal Optimizing an objective function.
• Can be applied to goal predicate type of
  problems.
• BSAT with objective function number of clauses
  satisfied.
• Intuition Always move to a better state

2
Some Hill-Climbing Algos

• Start State empty state or random state or
  special state
• Until (no improvement)
• Steepest Ascent find best successor
• OR (greedy) select first improving successor
• Go to that successor
• Repeat the above process some number of times
  (Restarts).
• Can be done with partial solutions or full
  solutions.

3
Local (Iterative) Improving

• Initial state full candidate solution
• Random or nearest neighbor selection
• Greedy hill-climbing
• if up, do it
• if flat, probabilistically decide to accept move
• Not necessary for homework
• Otherwise need to limit number of flat moves
• if down, dont do it
• We are gradually expanding the possible moves.

4
TSP Notes

• Only use up moves
• It is not necessary to compute the entire tour
  lengths after an operator applies. Instead
  compute change in length- most of the path length
  stays the same.
• You keep applying the operators until no
  improvement occurs.

5
Uncross Problem