6533b821fe1ef96bd127b75c
RESEARCH PRODUCT
Optimization problem in inductive inference
Andris Ambainissubject
CombinatoricsOptimization problemFinInductive probabilitySubroutineTotal functionFunction (mathematics)Inductive reasoningUpper and lower boundsMathematicsdescription
Algorithms recognizing to which of n classes some total function belongs are constructed (n > 2). In this construction strategies determining to which of two classes the function belongs are used as subroutines. Upper and lower bounds for number of necessary strategies are obtained in several models: FIN- and EX-identification and EX-identification with limited number of mindchanges. It is proved that in EX-identification it is necessary to use n(n−1)/2 strategies. In FIN-identification [3n/2 − 2] strategies are necessary and sufficient, in EX-identification with one mindchange- n log2n+o(n log2n) strategies.
year | journal | country | edition | language |
---|---|---|---|---|
1995-01-01 |