TinyTL: Reduce Activations, Not Trainable Parameters for Efficient On-Device Learning

TL;DR

TinyTL introduces a memory-efficient on-device learning method that freezes weights and learns only bias modules, significantly reducing memory usage while maintaining high accuracy.

Contribution

The paper proposes TinyTL, a novel approach that reduces memory consumption by freezing weights and learning lightweight bias modules, enabling efficient on-device adaptation.

Findings

Up to 6.5x memory savings with minimal accuracy loss

Significant accuracy improvements over last-layer fine-tuning

7.3-12.9x memory reduction with maintained accuracy

Abstract

On-device learning enables edge devices to continually adapt the AI models to new data, which requires a small memory footprint to fit the tight memory constraint of edge devices. Existing work solves this problem by reducing the number of trainable parameters. However, this doesn't directly translate to memory saving since the major bottleneck is the activations, not parameters. In this work, we present Tiny-Transfer-Learning (TinyTL) for memory-efficient on-device learning. TinyTL freezes the weights while only learns the bias modules, thus no need to store the intermediate activations. To maintain the adaptation capacity, we introduce a new memory-efficient bias module, the lite residual module, to refine the feature extractor by learning small residual feature maps adding only 3.8% memory overhead. Extensive experiments show that TinyTL significantly saves the memory (up to 6.5x) with little accuracy loss compared to fine-tuning the full network. Compared to fine-tuning the last layer, TinyTL provides significant accuracy improvements (up to 34.1%) with little memory overhead. Furthermore, combined with feature extractor adaptation, TinyTL provides 7.3-12.9x memory saving without sacrificing accuracy compared to fine-tuning the full Inception-V3.