We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.NE

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo

Computer Science > Neural and Evolutionary Computing

Title: Coevolutionary intransitivity in games: A landscape analysis

Abstract: Intransitivity is supposed to be a main reason for deficits in coevolutionary progress and inheritable superiority. Besides, coevolutionary dynamics is characterized by interactions yielding subjective fitness, but aiming at solutions that are superior with respect to an objective measurement. Such an approximation of objective fitness may be, for instance, generalization performance. In the paper a link between rating-- and ranking--based measures of intransitivity and fitness landscapes that can address the dichotomy between subjective and objective fitness is explored. The approach is illustrated by numerical experiments involving a simple random game with continuously tunable degree of randomness.
Subjects: Neural and Evolutionary Computing (cs.NE); Populations and Evolution (q-bio.PE)
Journal reference: In: Applications of Evolutionary Computation - EvoApplications 2015, (Eds.: A. M. Mora, G. Squillero), Lecture Notes in Computer Science, Vol. 9028, Springer-Verlag, Berlin, 2015, 869-881
Cite as: arXiv:1501.04010 [cs.NE]
  (or arXiv:1501.04010v1 [cs.NE] for this version)

Submission history

From: Hendrik Richter [view email]
[v1] Fri, 16 Jan 2015 15:22:19 GMT (2495kb,D)

Link back to: arXiv, form interface, contact.