Remembering Alex Fraser and explorations in learning without human expertise

Summary form only given. The article is divided into two parts. The first part will focus on Alex Fraser´s pioneering efforts in evolutionary computation dating from 1957-1970. Fraser, who passed away in July 2002, was one of the first to suggest and implement computer simulations of genetic systems that incorporated multiple individuals, recombination, mutation, inversion, and even a framework for placing evolutionary simulations in the context of adaptive purposive systems. His work from the 1950s and 1960s in Sydney, Australia was recognized by the IEEE Neural Networks Council with their pioneer award in evolutionary computation, presented at the 1999 Congress on Evolutionary Computation. The second part will change the focus to applying evolutionary computation to address problems for which human expertise may be insufficient or unavailable. In contrast to traditional artificial intelligence methods, which rely on capturing human expertise, evolutionary algorithms can be used to bootstrap knowledge from an environment over a series of generations. Results from the Blondie24 series of checkers experiments, in which a computer evolved a neural network that can compete at the level of human experts, will be reviewed and the implications of this research will be projected.