Brute-Force performs an exhaustive search, which enumerates all possible candidates for the solution to prove whether it satisfies the problem statement. Brute-force algorithms are conceptually simple, but usually suffer from exponential growing search space.
In 1949, Claude Shannon categorized brute-force minimax search as Type A strategy, which enumerates and minimaxes all leaf positions of a tree with an uniform depth, and concluded a machine operating according to the type A strategy would be both slow and a weak player [2]. Considering the branching factor potentiated by depth, Shannon favored the Type B strategy, a selective approach only searching "important" branches as the most promising.
However, with the advent of brute-force alpha-beta, and programs like Daly CP, Tech, Kaissa and Chess 4.5 in the early 70s, the era of the former dominating Type B programs drew to a close. Today most programs are closer to Type A, but have some characteristics of a Type B due to selectivity.
Backtracking
Backtracking discards large sets of incrementally build candidates to a solution, and "backtracks" a partial candidate as soon as it determines it cannot become member of the solution, for instance as demonstrated by the recursiveDe Bruijn Sequence Generator. Backtracking algorithms are not considered brute-force.
Jonathan Schaeffer, Paul Lu, Duane Szafron, Rob Lake (1993). A Re-examination of Brute-force Search. Intelligent Games: Planning and Learning. (AAAI 1993 Report FS9302, Proccedings of the AAAI Fall Symposiuem, eds. S. Epstein and R. Levinson), pp. 51-58, AAAI Press, Menol Park, CA. ISBN 0-929-28051-2. pdf
Miikka Maki-Uuro (1999). Limits to the brute force intelligence in computer chess. Proc. 1999 HeCSE Winter School, (ed. Tapio Elomaa), Report No. C-1999-1, Department of Computer Science, University of Helsinki[4]
Table of Contents
Search Algorithms
In 1949, Claude Shannon categorized brute-force minimax search as Type A strategy, which enumerates and minimaxes all leaf positions of a tree with an uniform depth, and concluded a machine operating according to the type A strategy would be both slow and a weak player [2]. Considering the branching factor potentiated by depth, Shannon favored the Type B strategy, a selective approach only searching "important" branches as the most promising.However, with the advent of brute-force alpha-beta, and programs like Daly CP, Tech, Kaissa and Chess 4.5 in the early 70s, the era of the former dominating Type B programs drew to a close. Today most programs are closer to Type A, but have some characteristics of a Type B due to selectivity.
Backtracking
Backtracking discards large sets of incrementally build candidates to a solution, and "backtracks" a partial candidate as soon as it determines it cannot become member of the solution, for instance as demonstrated by the recursive De Bruijn Sequence Generator. Backtracking algorithms are not considered brute-force.See also
Publications
1949
1970 ...
1980 ...
1990 ...
Forum Posts
External Links
References
What links here?
Up one Level