English

High-dimensional Neural Feature Design for Layer-wise Reduction of Training Cost

Machine Learning 2020-10-28 v2 Machine Learning

Abstract

We design a ReLU-based multilayer neural network by mapping the feature vectors to a higher dimensional space in every layer. We design the weight matrices in every layer to ensure a reduction of the training cost as the number of layers increases. Linear projection to the target in the higher dimensional space leads to a lower training cost if a convex cost is minimized. An 2\ell_2-norm convex constraint is used in the minimization to reduce the generalization error and avoid overfitting. The regularization hyperparameters of the network are derived analytically to guarantee a monotonic decrement of the training cost, and therefore, it eliminates the need for cross-validation to find the regularization hyperparameter in each layer. We show that the proposed architecture is norm-preserving and provides an invertible feature vector, and therefore, can be used to reduce the training cost of any other learning method which employs linear projection to estimate the target.

Keywords

Cite

@article{arxiv.2003.13058,
  title  = {High-dimensional Neural Feature Design for Layer-wise Reduction of Training Cost},
  author = {Alireza M. Javid and Arun Venkitaraman and Mikael Skoglund and Saikat Chatterjee},
  journal= {arXiv preprint arXiv:2003.13058},
  year   = {2020}
}

Comments

2020 EURASIP Journal on Advances in Signal Processing

R2 v1 2026-06-23T14:30:55.922Z