Advancing On-Device Neural Network Training with TinyPropv2: Dynamic, Sparse, and Efficient Backpropagation

TL;DR

TinyPropv2 is a novel sparse backpropagation algorithm optimized for low-power microcontrollers, enabling efficient on-device neural network training with minimal accuracy loss and reduced computational effort.

Contribution

The paper introduces TinyPropv2, a dynamic sparse backpropagation method that selectively skips training steps, significantly reducing computation while maintaining high accuracy.

Findings

Achieves near-parity with full training accuracy, with only ~1% drop.

Reduces computational effort to as low as 10% of full training.

Outperforms existing sparse training methods across multiple datasets.

Abstract

This study introduces TinyPropv2, an innovative algorithm optimized for on-device learning in deep neural networks, specifically designed for low-power microcontroller units. TinyPropv2 refines sparse backpropagation by dynamically adjusting the level of sparsity, including the ability to selectively skip training steps. This feature significantly lowers computational effort without substantially compromising accuracy. Our comprehensive evaluation across diverse datasets CIFAR 10, CIFAR100, Flower, Food, Speech Command, MNIST, HAR, and DCASE2020 reveals that TinyPropv2 achieves near-parity with full training methods, with an average accuracy drop of only around 1 percent in most cases. For instance, against full training, TinyPropv2's accuracy drop is minimal, for example, only 0.82 percent on CIFAR 10 and 1.07 percent on CIFAR100. In terms of computational effort, TinyPropv2 shows a marked reduction, requiring as little as 10 percent of the computational effort needed for full training in some scenarios, and consistently outperforms other sparse training methodologies. These findings underscore TinyPropv2's capacity to efficiently manage computational resources while maintaining high accuracy, positioning it as an advantageous solution for advanced embedded device applications in the IoT ecosystem.