Eurisko

Eurisko

Eurisko is a transformational system developed by Lenat in 1983 to address problems in artificial intelligence. It is designed to change its own heuristics and goals in order to make interesting discoveries in various fields. Eurisko is an example of transformational creativity, which involves transforming some dimension(s) of the conceptual space to generate new ideas that could not have arisen before. It has a way to evaluate rules to determine their contribution to a desirable outcome.From: Contemporary Artificial Intelligence [2019], Towards transformational creation of novel songs [2019], Risks of Artificial Intelligence [2019], A Creativity Support System Based on Causal Mapping [2018], Artificial Superintelligence [2019]

Unsupervised learning and reinforcement learning

View Chapter

Purchase Book

Published in Richard E. Neapolitan, Xia Jiang, Artificial Intelligence, 2018

Richard E. Neapolitan, Xia Jiang

In 1977, Doug Lenat [Davis and Lenat, 1982] developed Automated Mathematician (AM). AM was empowered with a large body of heuristics and by modifying LISP programs was able to discover the natural numbers, arithmetic, and several elementary concepts in number theory. However, it failed to discover anything advanced. Lenat [1983] developed EURISKO to address some of the problems in AM. EURISKO tries to learn new heuristics. Lenat and Brown [1984] discuss why AM and EURISKO appeared to work. Other efforts at automatic discovery of mathematical concepts appear in [Sims, 1987] and [Cotton et al., 2000].

Towards transformational creation of novel songs

View Article

Journal Information

Published in Connection Science, 2019

Jukka M. Toivanen, Matti Järvisalo, Olli Alm, Dan Ventura, Martti Vainio, Hannu Toivonen

Early examples of transformational systems in the field of artificial intelligence include Lenat's Automated Mathematician and Eurisko (Lenat, 1983; Lenat & Brown, 1984). They both are discovery systems consisting of heuristic rules, including heuristics about changing their own heuristics. They are clearly transformational in the sense of Boden (2004) and Wiggins (2006) as they manipulate their own search space. Automated Mathematician and Eurisko have inspired several successors; for instance, the HR system (Colton, 2001; Colton, Bundy, & Walsh, 1999) automatically forms mathematical theories using a higher level theory that contains concepts and conjectures about the lower level concepts and conjectures.

Explore chapters and articles related to this topic

Unsupervised learning and reinforcement learning

Towards transformational creation of novel songs