Time and Statistical Complexity of Proposed Evolutionary Algorithm in Artificial Neural Networks

Abstract

The important issue in Evolutionary Algorithms (EAs) analysis, is time-complexity. Here to obtain the mean hitting time of EA the concept of take-overtime is considered. The time complexity of the EA such as the takeover time is considered, i.e. the concept of the takeover time is generalized rather than a selection of operator alone. This generalization is applied to benchmark problems like N-Bit parity. For various input sizes N, the time complexity in terms of number of generations is estimated. An empirical model is also generated for proposed EA using statistical tool.