Advances in convolutional neural networks and their applications in atmospheric science

doi:10.11755/j.issn.1006-7639-2024-05-0719

Abstract

Abstract:

With the advancement of computer technology and big data, convolutional neural networks of the deep learning have become the mainstream technology for processing large-scale data with grid structure, especially in the field of computer vision. Convolutional neural networks have also been gradually applied in the field of atmospheric science to process multi-angle and multi-scale meteorological data. This paper reviews the progress of convolutional neural networks and their applications in atmospheric science, the conclusions are as following. Through the optimization of network depth and width and magnitude compression, the accuracy and efficiency of convolutional neural networks have been significantly improved, and they have become the mainstream technology for computer vision tasks. The convolutional neural network can process meteorological data efficiently, and has been applied in meteorological target recognition, extreme event detection, numerical model improvement and drought weather event prediction, etc., showing a good application prospect. The application of convolutional neural networks in atmospheric science is still in the exploratory stage, and faces challenges such as the complexity of meteorological data, the need for improvement of model structure and poor interpretability, so further research is needed to promote its development.

Key words: deep learning, atmospheric science, convolutional neural network, image recognition, improvements in numerical weather prediction

摘要：

随着计算机技术和大数据的进步，深度学习尤其是卷积神经网络已成为处理网格结构大规模数据的主流技术，特别是在计算机视觉领域。卷积神经网络也开始应用于大气科学领域，针对多角度、多尺度的气象数据进行处理。本文综述了卷积神经网络及其在大气科学中的应用进展，总结如下：通过网络深度、宽度的优化和量级压缩，卷积神经网络的准确率和效率显著提升，成为计算机视觉任务的主流技术；卷积神经网络能高效处理气象数据，已应用于气象目标识别、极端事件检测、数值模式改进及干旱气象事件预报等方面，显示出良好的应用前景；卷积神经网络在大气科学中的应用尚处于探索阶段，且面临气象数据复杂、模型结构改进需求和可解释性差等挑战，需深入研究以推动其发展。

关键词: 深度学习, 大气科学, 卷积神经网络, 图像识别, 数值模式改进

CLC Number:

P409

MA Minjin, CHEN Ran, CAO Yidan, ZHANG Xingyu, LI Yuebin. Advances in convolutional neural networks and their applications in atmospheric science[J]. Journal of Arid Meteorology, 2024, 42(5): 719-733.

马敏劲, 陈然, 曹译丹, 张星宇, 李岳彬. 卷积神经网络研究进展及其在大气科学中的应用[J]. 干旱气象, 2024, 42(5): 719-733.

Figures/Tables 12

Fig.1 Development of CNN in the 21st century

Fig.2 The basic structure of convolutional neural networks

Fig.3 The 2D single-channel convolution

Fig.4 Three common activation function images

Fig.5 Max pooling

Fig.6 Fully connected layer（two layers）

Fig.7 Neural network training process

Fig.8 Comparison of SGD （a） and SGDM （b）（from Yu et al.， 2014）

Fig.9 Differences in the way of convolution layers connections in ConvLSTM （a） and TrajGRU （b）（from Shi et al.， 2017b）

Fig.10 Reconstruction process of three channels of sky image （from Yao et al.， 2022）

Fig.11 Two-way mutual feedback coupled flow chart combining data assimilation and deep learning （from Brajard et al.， 2020）

Fig.12 Workflow diagram of the SPCAM （a） and NNCAM （b）（from Wang et al.， 2022）

References 95

[1]	曹春燕, 陈元昭, 刘东华, 等, 2015. 光流法及其在临近预报中的应用[J]. 气象学报, 73(3): 471-480.
[2]	陈锦鹏, 冯业荣, 黄奕丹, 等, 2024. 利用机器学习模拟湿物理参数化方案[J]. 气象学报, 82(1): 113-126.
[3]	陈锦鹏, 冯业荣, 蒙伟光, 等, 2021. 基于卷积神经网络的逐时降水预报订正方法研究[J]. 气象, 47(1): 60-70.
[4]	邓建国, 张素兰, 张继福, 等, 2020. 监督学习中的损失函数及应用研究[J]. 大数据, 6(1): 60-80. DOI
[5]	樊许磊, 2022. 用于资料同化预处理的卷积神经网络亮温重映射技术[D]. 武汉: 华中科技大学.
[6]	洪思弟, 赖绍钧, 林志玮, 等, 2020. 结合深度卷积网络及光学图像的降雨强度识别[J]. 福建农林大学学报:自然科学版, 49(4): 567-576.
[7]	刘雅忱, 2020. 人工智能下深度学习在气象预报中应用综述[J]. 计算机产品与流通(11),121.
[8]	刘亚伟, 郭辰婧, 赵绘萍, 2021. 卷积神经网络的发展及其在肿瘤影像学中的应用[J]. 吉林医药学院学报, 42(4): 290-292.
[9]	孙书魁, 范菁, 孙中强, 等, 2024. 基于深度学习的图像数据增强研究综述[J]. 计算机科学, 51(1): 150-167. DOI
[10]	魏森涛, 王澄海, 张飞民, 等, 2023. 基于土壤温、湿度记忆性的土壤湿度预测方法研究[J]. 干旱气象, 41(5): 783-791. DOI
[11]	夏旻, 申茂阳, 王舰锋, 等, 2018. 基于卷积神经网络的卫星云图云量计算[J]. 系统仿真学报, 30(5): 1 623-1 630.
[12]	袁凯, 庞晶, 李武阶, 等, 2023. 深度学习模型在2021年汛期武汉市雷达回波临近预报中的应用评估[J]. 干旱气象, 41(1): 173-185. DOI
[13]	袁雅涵, 王烁, 王文青, 等, 2023. 基于深度学习的积层混合云对流泡降水粒子特征研究[J]. 干旱气象, 41(6): 933-943.
[14]	BADRINARAYANAN V, KENDALL A, CIPOLLA R, 2017. SegNet: A deep convolutional encoder-decoder architecture for image segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 39(12): 2 481-2 495.
[15]	BAÑO-MEDINA J, MANZANAS R, GUTIÉRREZ J M, 2020. Configuration and intercomparison of deep learning neural models for statistical downscaling[J]. Geoscientific Model Development, 13(4): 2 109-2 124.
[16]	BARRON J T, 2017. Continuously differentiable exponential linear units[J/OL]. (2017-04-24)[2024-04-17]. https://arxiv.org/abs/1704.07483.
[17]	BOEING G, 2016. Visual analysis of nonlinear dynamical systems: Chaos, fractals, self-similarity and the limits of prediction[J]. Systems, 4(4): 37. DOI: 10.3390/systems4040037.
[18]	BRAJARD J, CARRASSI A, BOCQUET M, et al, 2020. Combining data assimilation and machine learning to emulate a dynamical model from sparse and noisy observations: A case study with the Lorenz 96 model[J]. Journal of Computational Science, 44: 101171. DOI: 10.1016/j.jocs.2020.101171.
[19]	BYUN J, JUN C, KIM J, et al, 2023. Deep learning-based rainfall prediction using cloud image analysis[J]. IEEE Transactions on Geoscience and Remote Sensing, 61: 1-11. DOI: 10.1109/TGRS.2023.3263872.
[20]	CHEN L C, PAPANDREOU G, KOKKINOS I, et al, 2018. DeepLab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected CRFs[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 40(4): 834-848.
[21]	CLEVERT D A, UNTERTHINER T, HOCHREITER S, 2015. Fast and accurate deep network learning by exponential linear units (ELUs)[C/OL]// International Conference on Learning Representations (ICLR). Caribe Hilton, San Juan, Puerto Rico. (2015-11-23)[2024-04-17]. https://arxiv.org/abs/1511.07289.
[22]	COHEN T S, GEIGER M, KOHLER J, et al, 2018. Spherical CNNs[J/OL]. (2018-01-30)[2024-04-17]. https://arxiv.org/abs/1801.10130.
[23]	DAWOOD M, ASIF A, MINHAS F U A A, 2020. Deep-PHURIE: Deep learning based hurricane intensity estimation from infrared satellite imagery[J]. Neural Computing and Applications, 32(13): 9 009-9 017.
[24]	DOZAT T, 2016. Incorporating nesterov momentum into Adam[C]// Proceedings of the 4th International Conference on Learning Representations (ICLR). San Juan, Puerto Rico.
[25]	DUCHI J C, HAZAN E, SINGER Y, 2011. Adaptive subgradient methods for online learning and stochastic optimization[J]. Journal of Machine Learning Research, 12: 2 121-2 159.
[26]	ELHOSEINY M, HUANG S, ELGAMMAL A, 2015. Weather classification with deep convolutional neural networks[C]// IEEE. 2015 IEEE International Conference on Image Processing (ICIP). Quebec City, QC, Canada: IEEE: 3 349-3 353.
[27]	FENG C, WANG W D, TIAN Y, et al, 2017. Estimate air quality based on mobile crowd sensing and big data[C]// IEEE. 2017 IEEE 18th International Symposium on A World of Wireless, Mobile and Multimedia Networks (WoWMoM). Macau, China: IEEE: 1-9.
[28]	FUKUSHIMA K, 1980. Neocognitron: A self organizing neural network model for a mechanism of pattern recognition unaffected by shift in position[J]. Biological Cybernetics, 36(4): 193-202. DOI PMID
[29]	GAGNE D J, HAUPT S E, NYCHKA D W, et al, 2019. Interpretable deep learning for spatial analysis of severe hailstorms[J]. Monthly Weather Review, 147(8): 2 827-2 845.
[30]	GIYENKO A, PALVANOV A, CHO Y, 2018. Application of convolutional neural networks for visibility estimation of CCTV images[C]// IEEE. 2018 International Conference on Information Networking (ICOIN). Chiang Mai, Thailand: IEEE: 875-879.
[31]	GLOROT X, BORDES A, BENGIO Y, 2011. Deep sparse rectifier neural networks[J]. Journal of Machine Learning Research, 15: 315-323.
[32]	GRÖNQUIST P, YAO C Y, BEN-NUN T, et al, 2021. Deep learning for post-processing ensemble weather forecasts[J]. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 379(2194): 20200092. DOI: 10.1098/rsta.2020.0092.
[33]	HAM Y G, KIM J H, LUO J J, 2019. Deep learning for multi-year ENSO forecasts[J]. Nature, 573(7775): 568-572.
[34]	HAN L, CHEN M X, CHEN K K, et al, 2021. A deep learning method for bias correction of ECMWF 24-240 h forecasts[J]. Advances in Atmospheric Sciences, 38(9): 1 444-1 459.
[35]	HE K M, SUN J, TANG X O, 2011. Single image haze removal using dark channel prior[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 33(12): 2 341-2 353.
[36]	HE K M, ZHANG X Y, REN S Q, et al, 2015. Delving deep into rectifiers: Surpassing human-level performance on ImageNet classification[C]// IEEE. Proceedings of the 2015 IEEE International Conference on Computer Vision (ICCV). IEEE Computer Society: 1 026-1 034.
[37]	HE K M, ZHANG X Y, REN S Q, et al, 2016. Deep residual learning for image recognition[C]// IEEE. 2016 IEEE Conference on Computer Vision and Pattern Recognition(CVPR). Las Vegas, NV, USA: IEEE: 770-778.
[38]	HINTON G E, SALAKHUTDINOV R R, 2006. Reducing the dimensionality of data with neural networks[J]. Science, 313(5786): 504-507. DOI PMID
[39]	HINTON G E, SRIVASTAVA N, KRIZHEVSKY A, et al, 2012. Improving neural networks by preventing co-adaptation of feature detectors[J/OL]. (2012-07-03)[2024-07-09]. https://arxiv.org/abs/1207.0580.
[40]	HOWARD A G, ZHU M, CHEN B, et al, 2017. MobileNets: Efficient convolutional neural networks for mobile vision applications[J/OL]. (2017-04-17)[2024-04-17]. https://arxiv.org/abs/1704.04861.
[41]	HUANG G, LIU Z, VAN DER MAATEN L, et al, 2017. Densely connected convolutional networks[C]// IEEE. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Honolulu, HI, USA: IEEE: 2 261-2 269.
[42]	HUBEL D H, WIESEL T N, 1962. Receptive fields, binocular interaction and functional architecture in the cat’s visual cortex[J]. Journal of Physiology, 160(1): 106-154.
[43]	JIANG Y H, CHENG W, GAO F, et al, 2022. A cloud classification method based on a convolutional neural network for FY-4A satellites[J]. Remote Sensing, 14(10): 2314. DOI: 10.3390/rs14102314.
[44]	KINGMA D P, BA J, 2014. Adam: A method for stochastic optimization[C/OL]// International Conference on Learning Representations (ICLR). San Diego, CA, USA. (2014-12-22)[2024-04-17]. https://arxiv.org/abs/1412.6980.
[45]	KLAMBAUER G, UNTERTHINER T, MAYR A, et al, 2017. Self-Normalizing neural networks[C/OL]// Neural Information Processing Systems (NIPS). Long Beach, CA, USA. (2017-09-07)[2024-04-17]. https://arxiv.org/abs/1706.02515.
[46]	KOPP M, TUO Y, DISSE M, 2019. Fully automated snow depth measurements from time-lapse images applying a convolutional neural network[J]. Science of The Total Environment, 697: 134213. DOI: 10.1016/j.scitotenv.2019.134213.
[47]	KRIZHEVSKY A, SUTSKEVER I, HINTON G E, 2017. ImageNet classification with deep convolutional neural networks[J]. Communications of the ACM, 60(6): 84-90.
[48]	KURTH T, TREICHLER S, ROMERO J, et al, 2018. Exascale deep learning for climate analytics[C]// IEEE. SC18:International Conference for High Performance Computing, Networking, Storage and Analysis. Dallas, TX, USA: IEEE: 649-660.
[49]	LAGERQUIST R, MCGOVERN A, GAGNE D J, 2019. Deep learning for spatially explicit prediction of synoptic-scale fronts[J]. Weather and Forecasting, 34(4): 1 137-1 160.
[50]	LECUN Y, BENGIO Y, HINTON G, 2015. Deep learning[J]. Nature, 521(7553): 436-444.
[51]	LECUN Y, BOTTOU L, BENGIO Y, et al, 1998. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE, 86(11): 2 278-2 324.
[52]	LI Y, MOMEN M, 2021. Detection of weather events in optical satellite data using deep convolutional neural networks[J]. Remote Sensing Letters, 12(12): 1 227-1 237.
[53]	LIN M, CHEN Q, YAN S C, 2013. Network In Network[J/OL]. (2013-12-16)[2024-04-17]. https://arxiv.org/abs/1312.4400.
[54]	LIU Y J, RACAH E, PRABHAT, et al, 2016. Application of deep convolutional neural networks for detecting extreme weather in climate datasets[J/OL]. (2016-05-04)[2024-04-17]. https://arxiv.org/abs/1605.01156.
[55]	LIU Z M, WANG Y X, VAIDYA S, et al, 2024. KAN: Kolmogorov-Arnold Networks[J/OL]. (2024-06-16)[2024-04-17]. https://arxiv.org/abs/2404.19756.
[56]	LIU Z, LIN Y T, CAO Y, et al, 2021. Swin transformer: Hierarchical vision transformer using shifted windows[C]// IEEE. International Conference on Computer Vision (ICCV). Montreal, QC, Canada: IEEE: 10 012-10 022.
[57]	LONG J, SHELHAMER E, DARRELL T, 2015. Fully convolutional networks for semantic segmentation[C]// IEEE. 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Boston, MA, USA: IEEE: 3 431-3 440.
[58]	LU C W, LIN D, JIA J Y, et al, 2014. Two-class weather classification[C]// IEEE. 2014 IEEE Conference on Computer Vision and Pattern Recognition. Columbus, OH, USA: IEEE: 3 718-3 725.
[59]	MAAS A L, HANNUN A Y, NG A Y, 2013. Rectifier nonlinearities improve neural network acoustic models[C]// International Conference on Machine Learning (ICML). Atlanta, USA.
[60]	NESTEROV Y, 1983. A method for solving the convex programming problem with convergence rate O(1/k^2)[J]. Proceedings of the USSR Academy of Sciences, 269: 543-547.
[61]	PRADHAN R, AYGUN R S, MASKEY M, et al, 2018. Tropical cyclone intensity estimation using a deep convolutional neural network[J]. IEEE Transactions on Image Processing, 27(2): 692-702. DOI PMID
[62]	RACAH E, BECKHAM C, MAHARAJ T, et al, 2016. Semi-supervised detection of extreme weather events in large climate datasets[C/OL]// Neural Information Processing Systems (NIPS). Long Beach, CA, USA. (2016-12-07)[2024-04-17]. https://arxiv.org/abs/1612.02095v1.
[63]	RONNEBERGER O, FISCHER P, BROX T, 2015. U-net: convolutional networks for biomedical image segmentation[C]// Proceedings of the 18th International Conference on Medical Image Computing and Computer-Assisted Intervention. Munich, Germany: Springer International Publishing: 234-241.
[64]	RUDER S, 2016. An overview of gradient descent optimization algorithms[J/OL]. (2016-09-15)[2024-04-17]. https://arxiv.org/abs/1609.04747.
[65]	RUMELHART D E, HINTON G E, WILLIAMS R J, 1986. Learning representations by back-propagating errors[J]. Nature, 323: 533-536.
[66]	SABOUR S, FROSST N, HINTON G E, 2017. Dynamic Routing Between Capsules[C/OL]// Neural Information Processing Systems (NIPS). Long Beach, CA, USA. (2017-11-07)[2024-04-17]. https://arxiv.org/abs/1710.09829.
[67]	SCHER S, 2018. Toward data-driven weather and climate forecasting: Approximating a simple general circulation model with deep learning[J]. Geophysical Research Letters, 45(22): 12 616-12 622.
[68]	SCHER S, MESSORI G, 2018. Predicting weather forecast uncertainty with machine learning[J]. Quarterly Journal of the Royal Meteorological Society, 144(717): 2 830-2 841.
[69]	SCHER S, MESSORI G, 2019. Weather and climate forecasting with neural networks: using general circulation models (GCMs) with different complexity as a study ground[J]. Geoscientific Model Development, 12(7): 2 797-2 809.
[70]	SCHULTZ M G, BETANCOURT C, GONG B, et al, 2021. Can deep learning beat numerical weather prediction?[J] Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 379(2194): 20200097. DOI: 10.1098/rsta.2020.0097.
[71]	SELVARAJU R R, COGSWELL M, DAS A, et al, 2020. Grad-CAM: Visual explanations from deep networks via gradient-based localization[J]. International Journal of Computer Vision, 128(2): 336-359.
[72]	SHI C Z, WANG C H, WANG Y, et al, 2017a. Deep convolutional activations-based features for ground-based cloud classification[J]. IEEE Geoscience and Remote Sensing Letters, 14(6): 816-820.
[73]	SHI X J, CHEN Z R, WANG H, et al, 2015. Convolutional LSTM network: A machine learning approach for precipitation nowcasting[C]// Proceedings of the 28th International Conference on Neural Information Processing Systems. Montreal, Canada: MIT Press: 802-810.
[74]	SHI X J, GAO Z H, LAUSEN L, et al, 2017b. Deep learning for precipitation nowcasting: A benchmark and a new model[C]// Proceedings of the 31st International Conference on Neural Information Processing Systems. Curran Associates Inc., Long Beach, California, USA: 5 622-5 632.
[75]	SINHA N K, GRISCIK M P, 1971. A stochastic approximation method[J]. IEEE Transactions on Systems, Man, and Cybernetics, SMC-1(4): 338-344.
[76]	SPRINGENBERG J T, DOSOVITSKIY A, BROX T, et al, 2014. Striving for simplicity: The all convolutional net[C/OL]// International Conference on Learning Representations (ICLR). San Diego, CA, USA. (2014-12-21)[2024-04-17]. https://arxiv.org/abs/1412.6806.
[77]	SZEGEDY C, LIU W, JIA Y Q, et al, 2015. Going deeper with convolutions[C]// IEEE. 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Boston, MA, USA: IEEE: 1-9.
[78]	TIELEMAN T, HINTON G, 2012. Lecture 6.5-rmsprop: divide the gradient by a running average of its recent magnitude. COURSERA: Neural networks for machine learning, 4(2): 26-31.
[79]	VANDAL T, KODRA E, GANGULY S, et al, 2017. DeepSD: Generating high resolution climate change projections through single image super-resolution[C]// 27th International Joint Conference on Artificial Intelligence (IJCAI). Halifax, NS, Canada, 1703.03126. DOI: 10.48550/arXiv.1703.03126.
[80]	WANG L Y, TAN Z M, 2023. Deep learning parameterization of the tropical cyclone boundary layer[J]. Journal of Advances in Modeling Earth Systems, 15(1): e2022MS003034. DOI: 10.1029/2022MS003034.
[81]	WANG X, HAN Y L, XUE W, et al, 2022. Stable climate simulations using a realistic general circulation model with neural network parameterizations for atmospheric moist physics and radiation processes[J]. Geoscientific Model Development, 15(9): 3 923-3 940.
[82]	WEYN J A, DURRAN D R, CARUANA R, 2019. Can machines learn to predict weather? Using deep learning to predict gridded 500-hPa geopotential height from historical weather data[J]. Journal of Advances in Modeling Earth Systems, 11(8): 2 680-2 693.
[83]	XIE S N, GIRSHICK R, DOLLAR P, et al, 2017. Aggregated residual transformations for deep neural networks[C]// IEEE. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Honolulu, HI: IEEE: 5 987-5 995.
[84]	XU B, WANG N Y, CHEN T Q, et al, 2015. Empirical evaluation of rectified activations in convolutional network[C]// International Conference on Machine Learning (ICML). Lille, France. DOI: 10.48550/arXiv.1505.00853.
[85]	YAO M Y, TAO D, WANG J T, et al, 2022. MARVAir: Meteorology augmented residual-based visual approach for crowdsourcing air quality inference[J]. IEEE Transactions on Instrumentation and Measurement, 71: 1-10.
[86]	YU D J, WANG H L, CHEN P Q, et al, 2014. Mixed pooling for convolutional neural networks[C]// The 9th International Conference on Rough Sets and Knowledge Technology. Shanghai, China. DOI: 10.1007/978-3-319-11740-9_34.
[87]	ZEILER M D, FERGUS R, 2013. Stochastic pooling for regularization of deep convolutional neural networks[C]// 1st International Conference on Learning Representations (ICLR). Scottsdale, United States. DOI: 10.48550/arXiv.1301.3557.
[88]	ZEILER M D, FERGUS R, 2014. Visualizing and understanding convolutional networks[C]// European Conference on Computer Vision (ECCV). Cham: Springer International Publishing. DOI: 10.1007/978-3-319-10590-1_53.
[89]	ZHANG J L, LIU P, ZHANG F, et al, 2018. CloudNet: Ground-based cloud classification with deep convolutional neural network[J]. Geophysical Research Letters, 45(16): 8 665-8 672.
[90]	ZHANG X Y, ZHOU X Y, LIN M X, et al, 2018. ShuffleNet: An extremely efficient convolutional neural network for mobile devices[C]// IEEE. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Salt Lake City, UT: IEEE: 6 848-6 856.
[91]	ZHAO B, LI X L, LU X Q, et al, 2018. A CNN-RNN architecture for multi-label weather recognition[J]. Neurocomputing, 322: 47-57.
[92]	ZHAO X, WEI H K, WANG H, et al, 2019. 3D-CNN-based feature extraction of ground-based cloud images for direct normal irradiance prediction[J]. Solar Energy, 181: 510-518.
[93]	ZHOU B L, KHOSLA A, LAPEDRIZA À, et al, 2016. Learning deep features for discriminative localization[C]// IEEE. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Las Vegas, NV, USA: IEEE: 2 921-2 929.
[94]	ZHOU K H, ZHENG Y G, LI B, et al, 2019. Forecasting different types of convective weather: A deep learning approach[J]. Journal of Meteorological Research, 33(5): 797-809.
[95]	ZOPH B, LE Q V, 2016. Neural architecture search with reinforcement learning[C/OL]// International Conference on Learning Representations (ICLR). Caribe Hilton, San Juan, Puerto Rico. (2016-11-05)[2024-04-17]. https://arxiv.org/abs/1611.01578.