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

Marcus Rüb, Axel Sikora, Daniel Mueller-Gritschneder

Publikation: Beitrag in Buch/Bericht/KonferenzbandKonferenzbeitragBegutachtung

Abstract

This study introduces EmbeddedTrain, an innovative algorithm optimized for on-device learning in deep neural networks, specifically designed for low-power microcontroller units. EmbeddedTrain refines sparse backpropagation by dynamically adjusting the level of sparity, 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 EmbeddedTrain achieves near-parity with full training methods, with an average accuracy drop of only around 1% in most cases. For instance, against full training, EmbeddedTrain's accuracy drop is minimal, for example, only 0.82% on CIFAR 10 and 1.07% on CIFAR100. In terms of computational effort, EmbeddedTrain shows a marked reduction, requiring as little as 10% of the computational effort needed for full training in some scenarios, and consistently outperforms other sparse training methodologies. These findings underscore EmbeddedTrain'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.

OriginalspracheEnglisch
Titel2024 International Joint Conference on Neural Networks, IJCNN 2024 - Proceedings
Herausgeber (Verlag)Institute of Electrical and Electronics Engineers Inc.
ISBN (elektronisch)9798350359312
DOIs
PublikationsstatusVeröffentlicht - 2024
Veranstaltung2024 International Joint Conference on Neural Networks, IJCNN 2024 - Yokohama, Japan
Dauer: 30 Juni 20245 Juli 2024

Publikationsreihe

NameProceedings of the International Joint Conference on Neural Networks

Konferenz

Konferenz2024 International Joint Conference on Neural Networks, IJCNN 2024
Land/GebietJapan
OrtYokohama
Zeitraum30/06/245/07/24

Fingerprint

Untersuchen Sie die Forschungsthemen von „Advancing On-Device Neural Network Training with TinyPropv2: Dynamic, Sparse, and Efficient Backpropagation“. Zusammen bilden sie einen einzigartigen Fingerprint.

Dieses zitieren