We show that if a population of neural network agents is allowed to interact during learning, so as to arrive at a consensus solution to the learning problem, then they can implicitly achieve complexity regularization. We call this learning paradigm, the classification game. We characterize the game-theoretic equilibria of this system, and show how low-complexity equilibria get selected. The benefit of finding a low-complexity solution is better expected generalization. We demonstrate this benefit through experiments. © Springer-Verlag Berlin Heidelberg 2010.
CITATION STYLE
Swarup, S. (2010). The classification game: Complexity regularization through interaction. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 6069 LNAI, pp. 289–303). https://doi.org/10.1007/978-3-642-14962-7_19
Mendeley helps you to discover research relevant for your work.