We develop and apply new methods for signal enhancement in high energy physics. We use neural networks as an ostensive system capable of recognizing patterns in input collider data, even when the patterns are not well defined. We show that neural networks give better signal enhancement than traditional methods.; In addition, we develop a genetic algorithm that is self-adaptive to the parameter space and the stage of evolution. It can also embed other algorithms and, therefore, it is an extension of existing maximization algorithms.; We show that certain problems in high energy physics cannot be solved by using local maximization algorithms and that the genetic algorithm presented here can be applied to these cases as well. We illustrate one such case by using neural networks trained by our genetic algorithm.; We give four specific applications--Higgs decay, four-jet top decay, six-jet top decay and tau jets tagging--which show the validity of these techniques. |