LAB: Learnable Activation Binarizer for Binary Neural Networks

Master Thesis (2022)
Author(s)

S.T. Falkena (TU Delft - Electrical Engineering, Mathematics and Computer Science)

Contributor(s)

H Jamali-Rad – Mentor (TU Delft - Pattern Recognition and Bioinformatics)

Jan Gemert – Graduation committee member (TU Delft - Pattern Recognition and Bioinformatics)

George Iosifidis – Coach (TU Delft - Embedded Systems)

Faculty
Electrical Engineering, Mathematics and Computer Science
Copyright
© 2022 Sieger Falkena
More Info
expand_more
Publication Year
2022
Language
English
Copyright
© 2022 Sieger Falkena
Graduation Date
24-03-2022
Awarding Institution
Delft University of Technology
Programme
['Computer Engineering | Embedded Software']
Faculty
Electrical Engineering, Mathematics and Computer Science
Reuse Rights

Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.

Abstract

Binary Neural Networks (BNNs) are receiving an upsurge of attention for bringing power-hungry deep learning towards edge devices. The traditional wisdom in this space is to employ sign(.) for binarizing featuremaps. 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 performance 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

Files

Thesis_report_SF.pdf
(pdf | 5.98 Mb)
License info not available