Playing Billiards in Version Space
- 1 January 1997
- journal article
- Published by MIT Press in Neural Computation
- Vol. 9 (1) , 99-122
- https://doi.org/10.1162/neco.1997.9.1.99
Abstract
A ray-tracing method inspired by ergodic billiards is used to estimate the theoretically best decision rule for a given set of linear separable examples. For randomly distributed examples, the billiard estimate of the single Perceptron with best average generalization probability agrees with known analytic results, while for real-life classification problems, the generalization probability is consistently enhanced when compared to the maximal stability Perceptron.Keywords
All Related Versions
This publication has 16 references indexed in Scilit:
- Gradient descent learning in perceptrons: A review of its possibilitiesPhysical Review E, 1995
- On the convergence of interior-reflective Newton methods for nonlinear minimization subject to boundsMathematical Programming, 1994
- On the complexity of approximating the maximal inscribed ellipsoid for a polytopeMathematical Programming, 1993
- The AdaTron: An Adaptive Perceptron AlgorithmEurophysics Letters, 1989
- Analysis of hidden units in a layered network trained to classify sonar targetsNeural Networks, 1988
- Learning of correlated patterns in spin-glass networks by local learning rulesPhysical Review Letters, 1987
- Eigenvalues of the Laplacian in Two DimensionsSIAM Review, 1984
- Exact solution of the Schrodinger equation for a particle in a tetrahedral boxJournal of Physics A: General Physics, 1982
- Markov Partitions for dispersed billiardsCommunications in Mathematical Physics, 1980
- On the ergodic properties of nowhere dispersing billiardsCommunications in Mathematical Physics, 1979