John Holland was unusual in his ability to absorb the essence of other disciplines, articulate grand overarching principles, and then back them up with computational mechanisms and mathematics. Unlike most researchers, Holland moved seamlessly among these three modes of thinking, developing models that were years ahead of their time. A close reading of his work reveals the antecedents of many ideas prevalent in machine learning today (such as reinforcement learning in non-Markovian environments and active learning). His seminal genetic algorithm spawned the field of evolutionary computation, and his insights and wisdom helped define what are today referred to as the "sciences of complexity."
Adaptive Computation: The Multidisciplinary Legacy of John H. Holland
By Stephanie Forrest, Melanie Mitchell
Communications of the ACM, Vol. 59 No. 8, Pages 58-63
Via Complexity Digest