AutoLR: Layer-wise Pruning and Auto-tuning of Learning Rates in Fine-tuning of Deep Networks

TL;DR

AutoLR introduces a layer-wise pruning and auto-tuning method for learning rates in deep network fine-tuning, improving performance and reducing complexity by aligning layer-specific learning rates with their roles.

Contribution

It proposes a novel algorithm that combines layer-wise pruning with auto-tuning of learning rates, addressing limitations of uniform learning rates in fine-tuning.

Findings

Achieved state-of-the-art results on multiple image retrieval datasets.

Demonstrated improved fine-tuning performance with reduced network complexity.

Validated effectiveness through extensive experiments.

Abstract

Existing fine-tuning methods use a single learning rate over all layers. In this paper, first, we discuss that trends of layer-wise weight variations by fine-tuning using a single learning rate do not match the well-known notion that lower-level layers extract general features and higher-level layers extract specific features. Based on our discussion, we propose an algorithm that improves fine-tuning performance and reduces network complexity through layer-wise pruning and auto-tuning of layer-wise learning rates. The proposed algorithm has verified the effectiveness by achieving state-of-the-art performance on the image retrieval benchmark datasets (CUB-200, Cars-196, Stanford online product, and Inshop). Code is available at https://github.com/youngminPIL/AutoLR.