English

Committees of deep feedforward networks trained with few data

Computer Vision and Pattern Recognition 2014-06-24 v1 Neural and Evolutionary Computing

Abstract

Deep convolutional neural networks are known to give good results on image classification tasks. In this paper we present a method to improve the classification result by combining multiple such networks in a committee. We adopt the STL-10 dataset which has very few training examples and show that our method can achieve results that are better than the state of the art. The networks are trained layer-wise and no backpropagation is used. We also explore the effects of dataset augmentation by mirroring, rotation, and scaling.

Keywords

Cite

@article{arxiv.1406.5947,
  title  = {Committees of deep feedforward networks trained with few data},
  author = {Bogdan Miclut and Thomas Kaester and Thomas Martinetz and Erhardt Barth},
  journal= {arXiv preprint arXiv:1406.5947},
  year   = {2014}
}