Tuning evaluation functions by maximizing concordance
Theoretical Computer Science - Advances in computer games
Hi-index | 0.00 |
The ELF algorithm learns an evaluation function that maps an object that is described by a conjunction of discrete variables to a real value. This is accomplished by simultaneously constructing binary features and adjusting a real-valued weight for each of the constructed features. The algorithm determines how many features it needs and what they should be in a deterministic manner.