Task 001/002 - Neuro-inspired Algorithms and Theory

ABSTRACT: Despite their tremendous successes, convolutional neural networks (CNNs) incur high computational/storage costs and are vulnerable to adversarial perturbations. Recent works on robust model compression address these challenges by combining model compression techniques with adversarial training. But these methods are unable to improve throughput (frames-per-second) on real-life hardware while simultaneously preserving robustness to adversarial perturbations. To overcome this problem, we propose the method of Generalized Depthwise-Separable (GDWS) convolution-an efficient, universal, post-training approximation of a standard 2D convolution. GDWS dramatically improves the throughput of a standard pre-trained network on real-life hardware while preserving its robustness. Lastly, GDWS is scalable to large problem sizes since it operates on pre-trained models and doesn't require any additional training. We establish the optimality of GDWS as a 2D convolution approximator and present exact algorithms for constructing optimal GDWS convolutions under complexity and error constraints. We demonstrate the effectiveness of GDWS via extensive experiments on CIFAR-10, SVHN, and ImageNet datasets. Our code can be found at https://github.com/hsndbk4/GDWS.

 

BIO: Hassan Dbouk received the B.E. degree with High Distinction in 2017 from the Department of Electrical and Computer Engineering at the American University of Beirut, Lebanon. He received his M.S. degree in 2019 from the Department of Electrical and Computer Engineering at the University of Illinois at Urbana-Champaign, where he is currently pursuing his Ph.D. degree under the supervision of Professor Naresh Shanbhag. His research interests lie at the intersection of machine learning, circuits, and architecture. He is the recipient of the Rambus fellowship from the ECE department at the University of Illinois in 2020-2021.