Model Rubik's Cube: Twisting Resolution, Depth and Width for TinyNets

By Kai Han, Yunhe Wang, Qiulin Zhang, Wei Zhang, Chunjing Xu, Tong Zhang. 2020

111 citations recorded by the source corpus.

To obtain excellent deep neural architectures, a series of techniques are carefully designed in EfficientNets. The giant formula for simultaneously enlarging the resolution, depth and width provides us a Rubik's cube for neural networks. So that we can find networks with high efficiency and excellent performance by twisting the three dimensions. This paper aims to explore the twisting rules for obtaining deep neural networks with minimum model sizes and computational costs. Different from the network enlarging, we observe that resolution and depth are more important than width for tiny networks. Therefore, the original method, i.e., the compound scaling in EfficientNet is no longer suitable. To this end, we summarize a tiny formula for downsizing neural architectures through a series of smaller models derived from the EfficientNet-B0 with the FLOPs constraint. Experimental results on the ImageNet benchmark illustrate that our TinyNet performs much better than the smaller version of EfficientNets using the inversed giant formula. For instance, our TinyNet-E achieves a 59.9% Top-1 accuracy with only 24M FLOPs, which is about 1.9% higher than that of the previous best MobileNetV3 with similar computational cost. Code will be available at this https URL, and this https URL.

Subfields: Computer Graphics and Computer-Aided Design, Computer Vision and Pattern Recognition, Model Architectures, Efficiency & Compression

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Graphics and Computer-Aided Design

1. Dacheng Tao
2. Lu Yuan
3. Chunhua Shen
4. Wenyu Liu
5. Gang Yu
6. Zicheng Liu
7. Xinggang Wang
8. Dongdong Chen

In Computer Vision and Pattern Recognition

1. Dacheng Tao
2. Luc Van Gool
3. Chunhua Shen
4. Lu Yuan
5. Ming-Hsuan Yang
6. Jiashi Feng
7. Xiaogang Wang
8. Alan L. Yuille

In Model Architectures

1. Lu Yuan
2. Xiyang Dai
3. Shiji Song
4. Dacheng Tao
5. Gao Huang
6. Dongdong Chen
7. Zicheng Liu
8. Yunhe Wang

In Efficiency & Compression

1. Shiji Song
2. Yunhe Wang
3. Gao Huang
4. Kai Han
5. Yulin Wang
6. Le Yang
7. Yizeng Han
8. Honghui Wang

Citations in the corpus

• Dynamic Neural Networks: A Survey (2021, background): “neously has attracted researches in both static [193], [194] and dynamic networks [195], [196], [197], [198].”
• Run, Don't Walk: Chasing Higher FLOPS for Faster Neural Networks (2023, methodology): “They have been widely adopted in mobile/edge-oriented networks, such as MobileNets [20, 21, 47], ShuffleNets [40, 76], GhostNet [13], EfficientNets [54, 55], TinyNet [14], Xception [8], CondenseNet [23, 70], TVConv [4], MnasNet [53], and FBNet [67].”
• TopFormer: Token Pyramid Transformer for Mobile Semantic Segmentation (2022, background): “EfficientNet [38, 39] and TinyNet [14] study the compound scaling of depth, width and resolution.”
• A Survey on Vision Transformer (2020, methodology): “As a contrast, the lightweight CNN model GhostNet [44, 45] can achieve similar performance with only about 600M FLOPs.”
• Avoiding Overfitting: A Survey on Regularization Methods for Convolutional Neural Networks (2022, background): “It is possible to verify that in the TinyNet work [15].”
• Mobile-Former: Bridging MobileNet and Transformer (2021, background): “EfficientNet [30,32] and TinyNet [13] study the compound scaling of depth, width and resolution.”