Evolving Innovative Solutions™ - Since 1993
Natural Selection, Inc. personnel pioneered the use of computational intelligence methods for gaming. Dr. George Burgin and Dr. Lawrence Fogel teamed in the 1960s to develop the Adaptive Maneuvering Logic (AML) for use in an advanced NASA flight simulator for combat pilot training. The AML made use of evolutionary programming to develop an intelligently interactive opponent for human pilots and reached a level of proficiency that routinely defeated trained pilots. Dr. Burgin also pioneered the use of evolutionary learning on nonzero-sum games such as the coordination, trust, bargaining, and prisoner's dilemma games.
From 1997 to 2006, Natural Selection, Inc. supported efforts to evolve neural networks to learn to play checkers and chess. The results of these efforts were highly successful, leading to the Blondie24 checkers program, described in the book Blondie24: Playing at the Edge of AI, and the Blondie25 chess program. Blondie24 was ranked in the top 500 of 120,000 people at zone.com in checkers in 1999. The Blondie25 chess program earned wins over Fritz 8.0, which was a program ranked in the top 5 in the world at the time, and Blondie25 also was the first machine learning chess program to defeat a human master (US, nationally ranked). These hallmarks in computer science were featured in Nature, Le Monde, and The New York Times. The intellectual property developed in these efforts was spun out in a sister company, Digenetics, Inc., which holds patents in these areas and has supported the development of two recent games, Ultimate Soccer Boss and My Forever Puppy (on Facebook).
Natural Selection, Inc. personnel also have experience applying evolutionary computation for population biology modeling, including games such as the hawk-dove game and prisoner's dilemma game. They were the first to demonstrate the inadequacy of traditional evolutionary game theory for modeling populations with regards to equilibrium states when using finite population sizes. NSI staff also has experience with probabilistic computational models of coevolutionary learning, multi-agent systems, robotics, and evolutionary game theory.
© 2012-2013 Natural Selection, Inc. All Rights Reserved.