Abstract
Neural Architecture Search (NAS) aims to automatically produce network architectures suitable to specific tasks on given datasets. Unlike previous NAS strategies based on reinforcement learning, genetic algorithm, Bayesian optimization, and differential programming, we formulate the NAS task as a Max-Flow problem on search space consisting of Directed Acyclic Graph (DAG) and thus propose a novel NAS approach, called MF-NAS, which defines the search space and designs the search strategy in a fully graphic manner. In MF-NAS, parallel edges with capacities are induced by combining different operations, including skip connection, convolutions and pooling, and the weights and capacities of the parallel edges are updated iteratively during the search process. Moreover, we interpret MF-NAS from the perspective of non-parametric density estimation and show the relationship between the flow of a graph and the corresponding classification accuracy of a neural network architecture. We evaluate the competitive efficacy of our proposed MF-NAS across different datasets with different search spaces that are used in DARTS/ENAS and NAS-Bench-201.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Similar content being viewed by others
Notes
- 1.
If there are only the normal cells in the architecture, as NAS-Bench-201, then \(m^* = m_{normal}^*\).
- 2.
For the search space in ENAS and DARTS, we set \(N = 4\), \(M = 2\), \(K = 8\); for the search space in NAS-Bench-201, we set \(N = 3\) and \(K = 5\) without constraining M.
- 3.
Many architectures can get the same best reward.
- 4.
Precisely, MF-NAS uses the search space of DARTS.
References
Baker, B., Gupta, O., Naik, N., Raskar, R.: Designing neural network architectures using reinforcement learning. In: ICLR (2017)
Bein, W.W., Brucker, P., Tamir, A.: Minimum cost flow algorithms for series-parallel networks. Discrete Appl. Math. 10, 117–124 (1985)
Bender, G., Kindermans, P., Zoph, B., Vasudevan, V., Le, Q.V.: Understanding and simplifying one-shot architecture search. In: ICML (2018)
Bengio, E., Jain, M., Korablyov, M., Precup, D., Bengio, Y.: Flow network based generative models for non-iterative diverse candidate generation. In: NeurIPS (2021)
Bengio, Y., Deleu, T., Hu, E.J., Lahlou, S., Tiwari, M., Bengio, E.: GFlowNet foundations. ar**v preprint ar**v:2111.09266 (2021)
Bergstra, J., Bardenet, R., Bengio, Y., Kégl, B.: Algorithms for hyper-parameter optimization. In: NeurIPS (2011)
Bi, K., Hu, C., **e, L., Chen, X., Wei, L., Tian, Q.: Stabilizing DARTS with amended gradient estimation on architectural parameters. ar**v:1910.11831 (2019)
Bonilla, E.V., Chai, K.M.A., Williams, C.K.I.: Multi-task gaussian process prediction. In: NeurIPS (2007)
Chao, X., Mengting, H., Xueqi, H., Chun-Guang, L.: Automated search space and search strategy selection for AutoML. Pattern Recognit. 124, 108474 (2022)
Chen, X., Hsieh, C.J.: Stabilizing differentiable architecture search via perturbation-based regularization. In: ICLR (2020)
Chen, X., **e, L., Wu, J., Tian, Q.: Progressive differentiable architecture search: bridging the depth gap between search and evaluation. In: ICCV (2019)
Chu, X., Wang, X., Zhang, B., Lu, S., Wei, X., Yan, J.: Darts-: robustly step** out of performance collapse without indicators. In: ICLR (2021)
Coates, A., Ng, A.Y., Lee, H.: An analysis of single-layer networks in unsupervised feature learning. In: AISTATS (2011)
Dong, X., Yang, Y.: One-shot neural architecture search via self-evaluated template network. In: ICCV (2019)
Dong, X., Yang, Y.: Searching for a robust neural architecture in four GPU hours. In: CVPR (2019)
Dong, X., Yang, Y.: An algorithm-agnostic NAS benchmark. In: ICLR (2020)
Griffin, G., Holub, A., Perona, P.: Caltech-256 object category dataset (2007)
He, K., Zhang, X., Ren, S., Sun, J.: Deep residual learning for image recognition. In: CVPR (2016)
Kandasamy, K., Neiswanger, W., Schneider, J., Póczos, B., **ng, E.P.: Neural architecture search with Bayesian optimisation and optimal transport. In: NeurIPS (2018)
Krizhevsky, A., Hinton, G., et al.: Learning multiple layers of features from tiny images. Technical report, Citeseer (2009)
de Laroussilhe, Q., Jastrzkebski, S., Houlsby, N., Gesmundo, A.: Neural architecture search over a graph search space. CoRR (2018)
Li, G., Qian, G., Delgadillo, I.C., Muller, M., Thabet, A., Ghanem, B.: SGAS: sequential greedy architecture search. In: CVPR (2020)
Li, L., Talwalkar, A.: Random search and reproducibility for neural architecture search. In: UAI (2019)
Li, L., Jamieson, K.G., DeSalvo, G., Rostamizadeh, A., Talwalkar, A.: Hyperband: a novel bandit-based approach to hyperparameter optimization. J. Mach. Learn. Res., 185:1–185:52 (2017)
Liu, C., et al.: Progressive neural architecture search. In: Ferrari, V., Hebert, M., Sminchisescu, C., Weiss, Y. (eds.) ECCV 2018. LNCS, vol. 11205, pp. 19–35. Springer, Cham (2018). https://doi.org/10.1007/978-3-030-01246-5_2
Liu, H., Simonyan, K., Vinyals, O., Fernando, C., Kavukcuoglu, K.: Hierarchical representations for efficient architecture search. In: ICLR (2018)
Liu, H., Simonyan, K., Yang, Y.: DARTS: differentiable architecture search. In: ICLR (2019)
Mnih, V., et al.: Nature (2015)
Muresan, H., Oltean, M.: Fruit recognition from images using deep learning. Acta Universitatis Sapientiae Informatica (2018)
Nguyen, V., Le, T., Yamada, M., Osborne, M.A.: Optimal transport kernels for sequential and parallel neural architecture search. In: ICML (2021)
Nilsback, M., Zisserman, A.: Automated flower classification over a large number of classes. In: ICVGIP (2008)
Pham, H., Guan, M.Y., Zoph, B., Le, Q.V., Dean, J.: Efficient neural architecture search via parameter sharing. In: ICML (2018)
Real, E., Aggarwal, A., Huang, Y., Le, Q.V.: Regularized evolution for image classifier architecture search. In: AAAI (2019)
Russakovsky, O., et al.: ImageNet large scale visual recognition challenge. In: IJCV (2015)
Shi, H., Pi, R., Xu, H., Li, Z., Kwok, J., Zhang, T.: Bridging the gap between sample-based and one-shot neural architecture search with BONAS. In: NeurIPS (2020)
Smith, S.L., Kindermans, P., Ying, C., Le, Q.V.: Don’t decay the learning rate, increase the batch size. In: ICLR (2018)
Snoek, J., Larochelle, H., Adams, R.P.: Practical Bayesian optimization of machine learning algorithms. In: NeurIPS (2012)
Su, X., et al.: Prioritized architecture sampling with Monto-Carlo tree search. In: CVPR (2021)
Swersky, K., Snoek, J., Adams, R.P.: Multi-task Bayesian optimization. In: NeurIPS (2013)
Wang, L., Fonseca, R., Tian, Y.: Learning search space partition for black-box optimization using Monte Carlo tree search. In: NeurIPS (2020)
Wang, R., Cheng, M., Chen, X., Tang, X., Hsieh, C.J.: Rethinking architecture selection in differentiable NAS. In: ICLR (2021)
Wang, X., Lin, J., Zhao, J., Yang, X., Yan, J.: EAutoDet: efficient architecture search for object detection. In: Farinella, T. (ed.) ECCV 2022. LNCS, vol. 13680, pp. 668–684 (2022)
Wang, X., Xue, C., Yan, J., Yang, X., Hu, Y., Sun, K.: MergeNAS: merge operations into one for differentiable architecture search. In: IJCAI (2020)
West, D.B., et al.: Introduction to Graph Theory, vol. 2. Prentice Hall, Upper Saddle River (1996)
White, C., Neiswanger, W., Savani, Y.: Bananas: Bayesian optimization with neural architectures for neural architecture search. In: AAAI (2021)
**e, S., Kirillov, A., Girshick, R.B., He, K.: Exploring randomly wired neural networks for image recognition. In: ICCV (2019)
**e, S., Zheng, H., Liu, C., Lin, L.: SNAS: stochastic neural architecture search. In: ICLR (2019)
Xu, Y., et al.: PC-DARTS: partial channel connections for memory-efficient architecture search. In: ICLR (2019)
Xue, C., Wang, X., Yan, J., Hu, Y., Yang, X., Sun, K.: Rethinking Bi-level optimization in neural architecture search: a gibbs sampling perspective. In: AAAI (2021)
Zela, A., Elsken, T., Saikia, T., Marrakchi, Y., Brox, T., Hutter, F.: Understanding and robustifying differentiable architecture search. In: ICLR (2020)
Zhou, H., Yang, M., Wang, J., Pan, W.: BayesNAS: a Bayesian approach for neural architecture search. In: ICML (2019)
Zoph, B., Le, Q.V.: Neural architecture search with reinforcement learning. In: ICLR (2017)
Zoph, B., Vasudevan, V., Shlens, J., Le, Q.V.: Learning transferable architectures for scalable image recognition. In: CVPR (2018)
Acknowledgments
J. Yan is supported by the National Key Research and Development Program of China under grant 2020AAA0107600. C.-G. Li is supported by the National Natural Science Foundation of China under grant 61876022.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2022 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Xue, C., Wang, X., Yan, J., Li, CG. (2022). A Max-Flow Based Approach for Neural Architecture Search. In: Avidan, S., Brostow, G., Cissé, M., Farinella, G.M., Hassner, T. (eds) Computer Vision – ECCV 2022. ECCV 2022. Lecture Notes in Computer Science, vol 13680. Springer, Cham. https://doi.org/10.1007/978-3-031-20044-1_39
Download citation
DOI: https://doi.org/10.1007/978-3-031-20044-1_39
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-20043-4
Online ISBN: 978-3-031-20044-1
eBook Packages: Computer ScienceComputer Science (R0)