Loading [a11y]/accessibility-menu.js
Rollback Ensemble With Multiple Local Minima in Fine-Tuning Deep Learning Networks | IEEE Journals & Magazine | IEEE Xplore

Rollback Ensemble With Multiple Local Minima in Fine-Tuning Deep Learning Networks


Abstract:

Image retrieval is a challenging problem that requires learning generalized features enough to identify untrained classes, even with very few classwise training samples. ...Show More

Abstract:

Image retrieval is a challenging problem that requires learning generalized features enough to identify untrained classes, even with very few classwise training samples. In this article, to obtain generalized features further in learning retrieval data sets, we propose a novel fine-tuning method of pretrained deep networks. In the retrieval task, we discovered a phenomenon in which the loss reduction in fine-tuning deep networks is stagnated, even while weights are largely updated. To escape from the stagnated state, we propose a new fine-tuning strategy to roll back some of the weights to the pretrained values. The rollback scheme is observed to drive the learning path to a gentle basin that provides more generalized features than a sharp basin. In addition, we propose a multihead ensemble structure to create synergy among multiple local minima obtained by our rollback scheme. Experimental results show that the proposed learning method significantly improves generalization performance, achieving state-of-the-art performance on the Inshop and SOP data sets.
Published in: IEEE Transactions on Neural Networks and Learning Systems ( Volume: 33, Issue: 9, September 2022)
Page(s): 4648 - 4660
Date of Publication: 03 March 2021

ISSN Information:

PubMed ID: 33657002

Funding Agency:


References

References is not available for this document.