MAE-DET: Revisiting Maximum Entropy Principle in Zero-Shot NAS for Efficient Object Detection

Zhenhong Sun, Ming Lin, Xiuyu Sun, Zhiyu Tan, Hao Li, Rong Jin
Proceedings of the 39th International Conference on Machine Learning, PMLR 162:20810-20826, 2022.

Abstract

In object detection, the detection backbone consumes more than half of the overall inference cost. Recent researches attempt to reduce this cost by optimizing the backbone architecture with the help of Neural Architecture Search (NAS). However, existing NAS methods for object detection require hundreds to thousands of GPU hours of searching, making them impractical in fast-paced research and development. In this work, we propose a novel zero-shot NAS method to address this issue. The proposed method, named MAE-DET, automatically designs efficient detection backbones via the Maximum Entropy Principle without training network parameters, reducing the architecture design cost to nearly zero yet delivering the state-of-the-art (SOTA) performance. Under the hood, MAE-DET maximizes the differential entropy of detection backbones, leading to a better feature extractor for object detection under the same computational budgets. After merely one GPU day of fully automatic design, MAE-DET innovates SOTA detection backbones on multiple detection benchmark datasets with little human intervention. Comparing to ResNet-50 backbone, MAE-DET is $+2.0%$ better in mAP when using the same amount of FLOPs/parameters, and is $1.54$ times faster on NVIDIA V100 at the same mAP. Code and pre-trained models are available here (https://github.com/alibaba/lightweight-neural-architecture-search).

Cite this Paper

BibTeX
@InProceedings{pmlr-v162-sun22c, title = {{MAE}-{DET}: Revisiting Maximum Entropy Principle in Zero-Shot {NAS} for Efficient Object Detection}, author = {Sun, Zhenhong and Lin, Ming and Sun, Xiuyu and Tan, Zhiyu and Li, Hao and Jin, Rong}, booktitle = {Proceedings of the 39th International Conference on Machine Learning}, pages = {20810--20826}, year = {2022}, editor = {Chaudhuri, Kamalika and Jegelka, Stefanie and Song, Le and Szepesvari, Csaba and Niu, Gang and Sabato, Sivan}, volume = {162}, series = {Proceedings of Machine Learning Research}, month = {17--23 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v162/sun22c/sun22c.pdf}, url = {https://proceedings.mlr.press/v162/sun22c.html}, abstract = {In object detection, the detection backbone consumes more than half of the overall inference cost. Recent researches attempt to reduce this cost by optimizing the backbone architecture with the help of Neural Architecture Search (NAS). However, existing NAS methods for object detection require hundreds to thousands of GPU hours of searching, making them impractical in fast-paced research and development. In this work, we propose a novel zero-shot NAS method to address this issue. The proposed method, named MAE-DET, automatically designs efficient detection backbones via the Maximum Entropy Principle without training network parameters, reducing the architecture design cost to nearly zero yet delivering the state-of-the-art (SOTA) performance. Under the hood, MAE-DET maximizes the differential entropy of detection backbones, leading to a better feature extractor for object detection under the same computational budgets. After merely one GPU day of fully automatic design, MAE-DET innovates SOTA detection backbones on multiple detection benchmark datasets with little human intervention. Comparing to ResNet-50 backbone, MAE-DET is $+2.0%$ better in mAP when using the same amount of FLOPs/parameters, and is $1.54$ times faster on NVIDIA V100 at the same mAP. Code and pre-trained models are available here (https://github.com/alibaba/lightweight-neural-architecture-search).} }
Endnote
%0 Conference Paper %T MAE-DET: Revisiting Maximum Entropy Principle in Zero-Shot NAS for Efficient Object Detection %A Zhenhong Sun %A Ming Lin %A Xiuyu Sun %A Zhiyu Tan %A Hao Li %A Rong Jin %B Proceedings of the 39th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2022 %E Kamalika Chaudhuri %E Stefanie Jegelka %E Le Song %E Csaba Szepesvari %E Gang Niu %E Sivan Sabato %F pmlr-v162-sun22c %I PMLR %P 20810--20826 %U https://proceedings.mlr.press/v162/sun22c.html %V 162 %X In object detection, the detection backbone consumes more than half of the overall inference cost. Recent researches attempt to reduce this cost by optimizing the backbone architecture with the help of Neural Architecture Search (NAS). However, existing NAS methods for object detection require hundreds to thousands of GPU hours of searching, making them impractical in fast-paced research and development. In this work, we propose a novel zero-shot NAS method to address this issue. The proposed method, named MAE-DET, automatically designs efficient detection backbones via the Maximum Entropy Principle without training network parameters, reducing the architecture design cost to nearly zero yet delivering the state-of-the-art (SOTA) performance. Under the hood, MAE-DET maximizes the differential entropy of detection backbones, leading to a better feature extractor for object detection under the same computational budgets. After merely one GPU day of fully automatic design, MAE-DET innovates SOTA detection backbones on multiple detection benchmark datasets with little human intervention. Comparing to ResNet-50 backbone, MAE-DET is $+2.0%$ better in mAP when using the same amount of FLOPs/parameters, and is $1.54$ times faster on NVIDIA V100 at the same mAP. Code and pre-trained models are available here (https://github.com/alibaba/lightweight-neural-architecture-search).
APA
Sun, Z., Lin, M., Sun, X., Tan, Z., Li, H. & Jin, R.. (2022). MAE-DET: Revisiting Maximum Entropy Principle in Zero-Shot NAS for Efficient Object Detection. Proceedings of the 39th International Conference on Machine Learning, in Proceedings of Machine Learning Research 162:20810-20826 Available from https://proceedings.mlr.press/v162/sun22c.html.

Related Material