Adaptive Feature Ranking for Unsupervised Transfer Learning

TL;DR

This paper introduces an adaptive feature ranking method for transfer learning that improves RBM training across different image datasets by selecting relevant representations without relying on target domain specifics.

Contribution

It presents a novel, general algorithm for ranking features from RBMs for transfer learning, applicable in unsupervised settings and across various domains.

Findings

Statistically significant improvements in RBM training performance

Effective feature ranking method applicable to multiple datasets

Unsupervised transfer learning without target domain dependence

Abstract

Transfer Learning is concerned with the application of knowledge gained from solving a problem to a different but related problem domain. In this paper, we propose a method and efficient algorithm for ranking and selecting representations from a Restricted Boltzmann Machine trained on a source domain to be transferred onto a target domain. Experiments carried out using the MNIST, ICDAR and TiCC image datasets show that the proposed adaptive feature ranking and transfer learning method offers statistically significant improvements on the training of RBMs. Our method is general in that the knowledge chosen by the ranking function does not depend on its relation to any specific target domain, and it works with unsupervised learning and knowledge-based transfer.