TY - GEN
T1 - Learning Lightweight Neural Networks via Channel-Split Recurrent Convolution
AU - Wu, Guojun
AU - Zhang, Xin
AU - Zhang, Ziming
AU - Li, Yanhua
AU - Zhou, Xun
AU - Brinton, Christopher
AU - Liu, Zhenming
N1 - Publisher Copyright:
© 2023 IEEE.
PY - 2023
Y1 - 2023
N2 - Lightweight neural networks refer to deep networks with small numbers of parameters, which can be deployed in resource-limited hardware such as embedded systems. To learn such lightweight networks effectively and efficiently, in this paper we propose a novel convolutional layer, namely Channel-Split Recurrent Convolution (CSR-Conv), where we split the output channels to generate data sequences with length T as the input to the recurrent layers with shared weights. As a consequence, we can construct lightweight convolutional networks by simply replacing (some) linear convolutional layers with CSR-Conv layers. We prove that under mild conditions the model size decreases with the rate of Oleft( {frac{1}{{{T2}}}} right). Empirically we demonstrate the state-of-the-art performance using VGG-16, ResNet-50, ResNet-56, ResNet-110, DenseNet-40, MobileNet, and EfficientNet as backbone networks on CIFAR-10 and ImageNet. Codes can be found on https://github.com/tuaxon/CSR_Conv.
AB - Lightweight neural networks refer to deep networks with small numbers of parameters, which can be deployed in resource-limited hardware such as embedded systems. To learn such lightweight networks effectively and efficiently, in this paper we propose a novel convolutional layer, namely Channel-Split Recurrent Convolution (CSR-Conv), where we split the output channels to generate data sequences with length T as the input to the recurrent layers with shared weights. As a consequence, we can construct lightweight convolutional networks by simply replacing (some) linear convolutional layers with CSR-Conv layers. We prove that under mild conditions the model size decreases with the rate of Oleft( {frac{1}{{{T2}}}} right). Empirically we demonstrate the state-of-the-art performance using VGG-16, ResNet-50, ResNet-56, ResNet-110, DenseNet-40, MobileNet, and EfficientNet as backbone networks on CIFAR-10 and ImageNet. Codes can be found on https://github.com/tuaxon/CSR_Conv.
KW - Algorithms: Machine learning architectures
KW - and algorithms (including transfer)
KW - formulations
UR - http://www.scopus.com/inward/record.url?scp=85149012285&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85149012285&partnerID=8YFLogxK
U2 - 10.1109/WACV56688.2023.00385
DO - 10.1109/WACV56688.2023.00385
M3 - Conference contribution
AN - SCOPUS:85149012285
T3 - Proceedings - 2023 IEEE Winter Conference on Applications of Computer Vision, WACV 2023
SP - 3847
EP - 3857
BT - Proceedings - 2023 IEEE Winter Conference on Applications of Computer Vision, WACV 2023
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 23rd IEEE/CVF Winter Conference on Applications of Computer Vision, WACV 2023
Y2 - 3 January 2023 through 7 January 2023
ER -