IEEE Computational Intelligence Magazine - August 2021 - 32

2) Results on COCO
We apply Mask R-CNN [25] with the C4 backbone as the
detector. The rest of the settings are the same as those for the
PASCAL VOC experiment. In Table 4, we can see the results
on the COCO dataset with the C4 backbone. With the 2 #
schedule, FaUNAE is better than its ImageNet supervised
counterpart on all metrics. Due to the absent result of MoCo
v2 [11], we do not compare it with our FaUNAE. We run their
code for this comparison, which is even worse than v1. Also,
FaUNAE is better than ResNet50 trained with unsupervised
MoCo v1 [1].
V. Conclusions
We proposed a fast and unsupervised neural architecture evolution
(FaUNAE) to evolve an existing architecture manually
designed or searched for on one small dataset to a new architecture
on another larger dataset. The evolution is self-supervised,
where contrast loss is used as the evaluation metric in the
student-teacher framework. The evolution process is significantly
accelerated by eliminating the inferior or the least promising
operation. Experimental results demonstrate the
effectiveness of our unsupervised NAE for the downstream
applications, such as object recognition, object detection, and
instance segmentation. FaUNAE addresses core technical challenges
in improving the performance of unsupervised deep
learning to make it more scalable and applicable, which will
ultimately impact many meaningful applications.
Acknowledgments
This study was supported by Grant NO.2019JZZY011101
from the Key Research and Development Program of Shandong
Province to Dianmin Sun. This work was supported in
part by the National Natural Science Foundation of China
under Grant 62076016 and 61876015.
References
[1] K. He, H. Fan, Y. Wu, S. Xie, and R. Girshick, " Momentum contrast for unsupervised
visual representation learning, " in Proc. CVPR, 2020, pp. 9726-9735.
[2] B. Zoph and Q. V. Le, " Neural architecture search with reinforcement learning, " in
Proc. ICLR, Toulon, France, Apr. 24-26, 2017.
[3] B. Zoph, V. Vasudevan, J. Shlens, and Q. V. Le, " Learning transferable architectures
for scalable image recognition, " in Proc. CVPR, 2018, pp. 8697-8710.
[4] H. Cai, L. Zhu, and S. Han, " ProxylessNAS: Direct neural architecture search on
target task and hardware, " in Proc. ICLR, 2019.
[5] H. Cai, T. Chen, W. Zhang, Y. Yu, and J. Wang, " Efficient architecture search by
network transformation, " in Proc. AAAI, 2018, pp. 2787-2794.
[6] C. Liu, P. Dollár, K. He, R. Girshick, A. Yuille, and S. Xie, " Are labels necessary for
neural architecture search? " 2020, arXiv:2003.12056.
[7] C. Doersch, A. Gupta, and A. A. Efros, " Unsupervised visual representation learning
by context prediction, " in Proc. ICCV, 2015, pp. 1422-1430.
[8] S. Gidaris, P. Singh, and N. Komodakis, " Unsupervised representation learning by
predicting image rotations, " in Proc. ICLR, 2018.
[9] A. Kolesnikov, X. Zhai, and L. Beyer, " Revisiting self-supervised visual representation
learning, " in Proc. CVPR.
[10] R. Hadsell, S. Chopra, and Y. LeCun, " Dimensionality reduction by learning an
invariant mapping, " in Proc. CVPR, 2006.
[11] X. Chen, H. Fan, R. Girshick, and K. He, " Improved baselines with momentum
contrastive learning, " 2020, arXiv:2003.04297.
[12] T. Chen, S. Kornblith, M. Norouzi, and G. Hinton, " A simple framework for contrastive
learning of visual representations, " 2020, arXiv:2002.05709.
[13] E. Real, S. Moore, A. Selle, S. Saxena, Y. L. Suematsu, J. Tan, Q. Le, and A. Kurakin,
" Large-scale evolution of image classifiers, " in Proc. ICML, pp. 2902-2911.
[14] H. Liu, K. Simonyan, and Y. Yang, " Darts: Differentiable architecture search, " in
Proc. ICLR, 2019.
[15] H. Pham, M. Y. Guan, B. Zoph, Q. V. Le, and J. Dean, " Efficient neural architecture
search via parameter sharing, " in Proc. ICML, 2018, pp. 4092-4101.
[16] Y. Xu, L. Xie, X. Zhang, X. Chen, G.-J. Qi, Q. Tian, and H. Xiong, " Pc-darts: Partial
channel connections for memory-efficient architecture search, " in Proc. ICLR, 2020.
[17] H. Chen, B. Zhang, S. Xue, X. Gong, H. Liu, R. Ji, and D. Doermann, " Anti-bandit
neural architecture search for model defense, " 2020, arXiv:2008.00698.
[18] T. Lin, M. Maire, S. J. Belongie, J. Hays, P. Perona, D. Ramanan, P. Dollár, and C.
L. Zitnick, " Microsoft COCO: Common objects in context, " in Proc. ECCV, SpringerVerlag,
2014, vol. 8693, pp. 740-755.
[19] M. Everingham, L. V. Gool, C. K. I. Williams, J. M. Winn, and A. Zisserman, " The
pascal visual object classes (VOC) challenge, " IJCV, vol. 88, no. 2, pp. 303-338, 2010. doi:
10.1007/s11263-009-0275-4.
[20] R. B. Girshick, J. Donahue, T. Darrell, and J. Malik, " Rich feature hierarchies for
accurate object detection and semantic segmentation, " in Proc. CVPR, 2014, pp. 580-587.
[21] R. Girshick, " Fast R-CNN, " in Proc. ICCV, 2015, pp. 1440-1448.
[22] S. Ren, K. He, R. Girshick, and J. Sun, " Faster R-CNN: Towards real-time object
detection with region proposal networks, " in Proc. NeurIPS, 2015, pp. 91-99.
[23] J. Redmon, S. K. Divvala, R. B. Girshick, and A. Farhadi, " You only look once:
Unified, real-time object detection, " in Proc. CVPR, Las Vegas, NV, 2016, pp. 779-788.
[24] W. Liu, D. Anguelov, D. Erhan, C. Szegedy, S. E. Reed, C. Fu, and A. C. Berg,
" SSD: Single shot multibox detector, " in Proc. Comput. Vision - ECCV 2016 - 14th Eur.
Conf., Amsterdam, The Netherlands, Springer-Verlag, 2016, vol. 9905, pp. 21-37.
[25] K. He, G. Gkioxari, P. Dollár, and R. Girshick, " Mask R-CNN, " in Proc. ICCV,
2017, pp. 2980-2988.
[26] X. Wang, T. Kong, C. Shen, Y. Jiang, and L. Li, " SOLO: Segmenting objects by
locations, " CoRR, vol. abs/1912.04488, 2019.
[27] X. Wang, R. Zhang, T. Kong, L. Li, and C. Shen, " SOLOv2: Dynamic, faster and
stronger, " 2020, arXiv:2003.10152.
[28] D. Floreano, P. Durr, and C. Mattiussi, " Neuroevolution: from architectures to
learning, " Evol. Intell., vol. 1, pp. 47-62, 2008. doi: 10.1007/s12065-007-0002-4.
[29] K. O. Stanley and R. Miikkulainen, " Evolving neural networks through augmenting
topologies, " MIT Press Journals, 2002.
[30] E. Real, A. Aggarwal, Y. Huang, and Q. V. Le, " Regularized evolution for image
classifier architecture search, " in Proc. AAAI, 2019, pp. 4780-4789.
[31] R. Miikkulainen et al., " Evolving deep neural networks, " 2017, arXiv:1703.00548.
[32] H. Liu, K. Simonyan, O. Vinyals, C. Fernando, and K. Kavukcuoglu, " Hierarchical
representations for efficient architecture search, " in Proc. ICLR, 2018.
[33] A. Krizhevsky et al., " Learning multiple layers of features from tiny images, " 2009.
[34] K. He, X. Zhang, S. Ren, and J. Sun, " Deep residual learning for image recognition, "
in Proc. CVPR, 2016, pp. 770-778.
[35] H. Zhang et al., " Resnest: Split-attention networks, " 2020, arXiv:2004.08955.
[36] E. Real, A. Aggarwal, Y. Huang, and Q. V. Le, " Regularized evolution for image classifier
architecture search, " in Proc. AAAI, 2019, pp. 4780-4789. Doi: 10.1609/aaai.v33
i01.33014780.
[37] R. Hadsell, S. Chopra, and Y. LeCun, " Dimensionality reduction by learning an
invariant mapping, " in Proc. CVPR, 2006, pp. 1735-1742.
[38] Y. Tian, D. Krishnan, and P. Isola, " Contrastive representation distillation, " in Proc.
ICLR, 2020.
[39] A. v. d. Oord, Y. Li, and O. Vinyals, " Representation learning with contrastive predictive
coding, " 2018, arXiv:1807.03748.
[40] Z. Wu, Y. Xiong, S. X. Yu, and D. Lin, " Unsupervised feature learning via nonparametric
instance discrimination, " in Proc. CVPR, 2018, pp. 3733-3742.
[41] X. Wang and A. Gupta, " Unsupervised learning of visual representations using videos, "
in Proc. CVPR, 2015, pp. 2794-2802.
[42] R. D. Hjelm, A. Fedorov, S. Lavoie-Marchildon, K. Grewal, P. Bachman, A.
Trischler, and Y. Bengio, " Learning deep representations by mutual information estimation
and maximization, " in Proc. ICLR, 2019.
[43] A. Tarvainen and H. Valpola, " Mean teachers are better role models: Weight-averaged
consistency targets improve semi-supervised deep learning results, " in Proc. NIPS,
2017.
[44] P. Bachman, R. D. Hjelm, and W. Buchwalter, " Learning representations by maximizing
mutual information across views, " in Proc. NeurIPS, 2019, pp. 15,509-15,519.
[45] C. Zhuang, A. L. Zhai, and D. Yamins, " Local aggregation for unsupervised learning
of visual embeddings, " in Proc. ICCV, 2019, pp. 6001-6011.
[46] O. J. Hénaff, A. Srinivas, J. De Fauw, A. Razavi, C. Doersch, S. Eslami, and A. v. d.
Oord, " Data-efficient image recognition with contrastive predictive coding, " 2019,
arXiv:1905.09272.
[47] Y. Tian, D. Krishnan, and P. Isola, " Contrastive multiview coding, " 2019, arXiv:1906.05849.
[48]
Y. Wu, A. Kirillov, F. Massa, W.-Y. Lo, and R. Girshick, " Detectron2. " 2019.
https://github.com/facebookresearch/detectron2
32 IEEE COMPUTATIONAL INTELLIGENCE MAGAZINE | AUGUST 2021
https://www.github.com/facebookresearch/detectron2
IEEE Computational Intelligence Magazine - August 2021

Table of Contents for the Digital Edition of IEEE Computational Intelligence Magazine - August 2021
