Abstract
Binary Neural Networks (BNNs) are receiving an up-surge of attention for bringing power-hungry deep learning towards edge devices. The traditional wisdom in this space is to employ sign(.) for binarizing feature maps. We argue and illustrate that sign(.) is a uniqueness bottleneck, limiting information propagation throughout the network. To alleviate this, we propose to dispense sign(.), replacing it with a learnable activation binarizer (LAB), allowing the network to learn a fine-grained binarization kernel per layer - as opposed to global thresholding. LAB is a novel universal module that can seamlessly be integrated into existing architectures. To confirm this, we plug it into four seminal BNNs and show a considerable accuracy boost at the cost of tolerable increase in delay and complexity. Finally, we build an end-to-end BNN (coined as LAB-BNN) around LAB, and demonstrate that it achieves competitive performance on par with the state-of-the-art on ImageNet. Our code can be found in our repository: https://github.com/sfalkena/LAB.
Original language | English |
---|---|
Title of host publication | Proceedings of the 2023 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) |
Editors | Lisa O’Conner |
Place of Publication | Piscataway |
Publisher | IEEE |
Pages | 6414-6423 |
Number of pages | 10 |
ISBN (Electronic) | 978-1-6654-9346-8 |
ISBN (Print) | 978-1-6654-9347-5 |
DOIs | |
Publication status | Published - 2023 |
Event | WACV: 2023 IEEE Winter Conference on Applications of Computer Vision - Waikoloa, United States Duration: 2 Jan 2023 → 7 Jan 2023 |
Conference
Conference | WACV |
---|---|
Country/Territory | United States |
City | Waikoloa |
Period | 2/01/23 → 7/01/23 |
Bibliographical note
Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-careOtherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.