Learning from Minimum Entropy Queries in a Large Committee Machine
adap-org
2009-10-28 v1 凝聚态物理
适应与自组织系统
摘要
In supervised learning, the redundancy contained in random examples can be avoided by learning from queries. Using statistical mechanics, we study learning from minimum entropy queries in a large tree-committee machine. The generalization error decreases exponentially with the number of training examples, providing a significant improvement over the algebraic decay for random examples. The connection between entropy and generalization error in multi-layer networks is discussed, and a computationally cheap algorithm for constructing queries is suggested and analysed.
引用
@article{arxiv.adap-org/9604001,
title = {Learning from Minimum Entropy Queries in a Large Committee Machine},
author = {Peter Sollich},
journal= {arXiv preprint arXiv:adap-org/9604001},
year = {2009}
}
备注
4 pages, REVTeX, multicol, epsf, two postscript figures. To appear in Physical Review E (Rapid Communications)