Maintained by Difan Deng and Marius Lindauer.
The following list considers papers related to neural architecture search. It is by no means complete. If you miss a paper on the list, please let us know.
Please note that although NAS methods steadily improve, the quality of empirical evaluations in this field are still lagging behind compared to other areas in machine learning, AI and optimization. We would therefore like to share some best practices for empirical evaluations of NAS methods, which we believe will facilitate sustained and measurable progress in the field. If you are interested in a teaser, please read our blog post or directly jump to our checklist.
2021 |
|
| 1237. | Lin, Yi; Liu, Yanfei; Liu, Jingguang; Liu, Guocai; Ma, Kai; Zheng, Yefeng LE-NAS: Learning-based Ensemble with Neural Architecture Search for 3D Radiotherapy Dose Prediction Journal Article 2021. @article{Lin2106, title = {LE-NAS: Learning-based Ensemble with Neural Architecture Search for 3D Radiotherapy Dose Prediction}, author = {Yi Lin and Yanfei Liu and Jingguang Liu and Guocai Liu and Kai Ma and Yefeng Zheng}, url = {https://arxiv.org/pdf/2106.06733.pdf}, year = {2021}, date = {2021-07-01}, keywords = {}, pubstate = {published}, tppubtype = {article} } |
| 1236. | Dimanov, Daniel; Balaguer-Ballester, Emili; Singleton, Colin; Rostami, Shahin Moncae: Multi-Objective Neuroevolution of Convolutional Autoencoders Inproceedings 2nd Workshop on Neural Architecture Search at ICLR 2021, 2021. @inproceedings{Dimanov2021, title = {Moncae: Multi-Objective Neuroevolution of Convolutional Autoencoders}, author = {Daniel Dimanov and Emili Balaguer-Ballester and Colin Singleton and Shahin Rostami}, url = {https://www.researchgate.net/profile/Daniel-Dimanov-2/publication/352248797_MONCAE_MultiI-Objective_Neuroevolution_of_Convolutional_Autoencoders/links/60c0a43592851ca6f8d2d98b/MONCAE-MultiI-Objective-Neuroevolution-of-Convolutional-Autoencoders.pdf}, year = {2021}, date = {2021-07-01}, booktitle = {2nd Workshop on Neural Architecture Search at ICLR 2021}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } |
| 1235. | Jin, Yaochu; Wang, Handing; Sun, Chaoli Surrogate-Assisted Evolutionary Neural Architecture Search Inproceedings Data-Driven Evolutionary Optimization, pp. 373-387, 2021, ISBN: 978-3-030-74640-7. @inproceedings{Jin2021, title = {Surrogate-Assisted Evolutionary Neural Architecture Search}, author = {Yaochu Jin and Handing Wang and Chaoli Sun}, url = {https://link.springer.com/chapter/10.1007/978-3-030-74640-7_12}, isbn = {978-3-030-74640-7}, year = {2021}, date = {2021-06-29}, booktitle = {Data-Driven Evolutionary Optimization}, volume = {975}, pages = {373-387}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } |
| 1234. | Zhu, Wei; Ni, Yuan; Wang, Xiaoling; Xie, Guotong; Zhang, Fang Discovering Better Model Architectures for Medical Query Understanding Inproceedings Proceedings of NAACL HLT 2021: IndustryTrack Papers, pp. 230-237, 2021. @inproceedings{ZHU2021, title = {Discovering Better Model Architectures for Medical Query Understanding}, author = { Wei Zhu and Yuan Ni and Xiaoling Wang and Guotong Xie and Fang Zhang}, url = {https://www.aclweb.org/anthology/2021.naacl-industry.29.pdf}, year = {2021}, date = {2021-06-06}, booktitle = {Proceedings of NAACL HLT 2021: IndustryTrack Papers}, pages = {230-237}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } |
| 1233. | Dai, Xiaoliang; Wan, Alvin; Zhang, Peizhao; Wu, Bichen; He, Zijian; Wei, Zhen; Chen, Kan; Tian, Yuandong; Yu, Matthew; Vajda, Peter; Gonzalez, Joseph E FBNetV3: Joint Architecture-Recipe Search using Predictor Pretraining Incollection CVPR2021, 2021. @incollection{DaiCVPR2021, title = {FBNetV3: Joint Architecture-Recipe Search using Predictor Pretraining}, author = {Xiaoliang Dai and Alvin Wan and Peizhao Zhang and Bichen Wu and Zijian He and Zhen Wei and Kan Chen and Yuandong Tian and Matthew Yu and Peter Vajda and Joseph E. Gonzalez}, url = {https://openaccess.thecvf.com/content/CVPR2021/papers/Dai_FBNetV3_Joint_Architecture-Recipe_Search_Using_Predictor_Pretraining_CVPR_2021_paper.pdf}, year = {2021}, date = {2021-06-01}, booktitle = {CVPR2021}, keywords = {}, pubstate = {published}, tppubtype = {incollection} } |
| 1232. | Ma, Xiaohan; Si, Chang; Wang, Ying; Liu, Cheng; Zhang1, Lei NASA: Accelerating Neural Network Design with a NAS Processor Inproceedings 2021 ACM/IEEE 48th Annual International Symposium on Computer Architecture (ISCA), 2021. @inproceedings{MaISCA2021, title = {NASA: Accelerating Neural Network Design with a NAS Processor}, author = {Xiaohan Ma and Chang Si and Ying Wang and Cheng Liu and Lei Zhang1}, url = {https://conferences.computer.org/iscapub/pdfs/ISCA2021-4ghucdBnCWYB7ES2Pe4YdT/333300a790/333300a790.pdf}, year = {2021}, date = {2021-06-01}, booktitle = {2021 ACM/IEEE 48th Annual International Symposium on Computer Architecture (ISCA)}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } |
| 1231. | Liu, Shaoli; Zheng, Chengjian; Lu, Kaidi; Gao, Si; Wang, Ning; Wang, Bofei; Zhang, Diankai; Zhang, Xiaofeng; Xu, Tianyu EVSRNet: Efficient Video Super-Resolution With Neural Architecture Search Inproceedings Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, pp. 2480-2485, 2021. @inproceedings{Liu_2021_CVPR, title = {EVSRNet: Efficient Video Super-Resolution With Neural Architecture Search}, author = {Shaoli Liu and Chengjian Zheng and Kaidi Lu and Si Gao and Ning Wang and Bofei Wang and Diankai Zhang and Xiaofeng Zhang and Tianyu Xu}, url = {https://openaccess.thecvf.com/content/CVPR2021W/MAI/html/Liu_EVSRNet_Efficient_Video_Super-Resolution_With_Neural_Architecture_Search_CVPRW_2021_paper.html}, year = {2021}, date = {2021-06-01}, booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops}, pages = {2480-2485}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } |
| 1230. | Geraeinejad, V; Seenan, S; Modarressi, M; Daneshtalab, M RoCo-NAS : Robust and Compact Neural Architecture Search Inproceedings The international joint conference on neural networks IJCNN, 2021. @inproceedings{Geraeinejad2021, title = {RoCo-NAS : Robust and Compact Neural Architecture Search}, author = {V. Geraeinejad and S. Seenan and M. Modarressi and M. Daneshtalab}, url = {https://www.diva-portal.org/smash/record.jsf?pid=diva2%3A1558970&dswid=-1593}, year = {2021}, date = {2021-06-01}, booktitle = {The international joint conference on neural networks IJCNN}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } |
| 1229. | Tsokov, Stefan ; Lazarova, Milena ; Aleksieva-Petrova, Adelina An Evolutionary Approach to the Design of Convolutional Neural Networks for Human Activity Recognition Journal Article Indian Journal of Computer Science and Engineering, 2021. @article{Tsokov2021, title = {An Evolutionary Approach to the Design of Convolutional Neural Networks for Human Activity Recognition}, author = {Tsokov, Stefan and Lazarova, Milena and Aleksieva-Petrova, Adelina}, url = {http://www.ijcse.com/docs/INDJCSE21-12-02-145.pdf}, year = {2021}, date = {2021-06-01}, journal = {Indian Journal of Computer Science and Engineering}, keywords = {}, pubstate = {published}, tppubtype = {article} } |
| 1228. | Chen, Wuyang; Gong, Xinyu; Wang, Zhangyang Neural Architecture Search on ImageNet in Four GPU Hours: A Theoretically Inspired Perspective Inproceedings ICLR 2021, 2021. @inproceedings{chen2021neural, title = {Neural Architecture Search on ImageNet in Four GPU Hours: A Theoretically Inspired Perspective}, author = {Wuyang Chen and Xinyu Gong and Zhangyang Wang}, url = {https://arxiv.org/abs/2102.11535}, year = {2021}, date = {2021-05-04}, booktitle = {ICLR 2021}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } |
| 1227. | Anwar, Abrar Evolving Spiking Circuit Motifs Using Weight Agnostic Neural Networks Journal Article Proceedings of the AAAI Conference on Artificial Intelligence, 35 (18), pp. 15956-15957, 2021. @article{Anwar_2021, title = {Evolving Spiking Circuit Motifs Using Weight Agnostic Neural Networks}, author = {Abrar Anwar}, url = {https://ojs.aaai.org/index.php/AAAI/article/view/17974}, year = {2021}, date = {2021-05-01}, journal = {Proceedings of the AAAI Conference on Artificial Intelligence}, volume = {35}, number = {18}, pages = {15956-15957}, keywords = {}, pubstate = {published}, tppubtype = {article} } |
| 1226. | Wang, Xiaobo Teacher Guided Neural Architecture Search for Face Recognition Journal Article Proceedings of the AAAI Conference on Artificial Intelligence, 35 (4), pp. 2817-2825, 2021. @article{Wang_2021, title = {Teacher Guided Neural Architecture Search for Face Recognition}, author = {Xiaobo Wang}, url = {https://ojs.aaai.org/index.php/AAAI/article/view/16387}, year = {2021}, date = {2021-05-01}, journal = {Proceedings of the AAAI Conference on Artificial Intelligence}, volume = {35}, number = {4}, pages = {2817-2825}, keywords = {}, pubstate = {published}, tppubtype = {article} } |
| 1225. | Perenda, Erma; Rajendran, Sreeraj; Bovet, Gerome; Pollin, Sofie; Zheleva, Mariya Evolutionary Optimization of Residual Neural Network Architectures for Modulation Classification Miscellaneous 2021. @misc{perenda_rajendran_bovet_pollin_zheleva_2021, title = {Evolutionary Optimization of Residual Neural Network Architectures for Modulation Classification}, author = {Erma Perenda and Sreeraj Rajendran and Gerome Bovet and Sofie Pollin and Mariya Zheleva}, url = {https://www.techrxiv.org/articles/preprint/Evolutionary_Optimization_of_Residual_Neural_Network_Architectures_for_Modulation_Classification/14528778/1}, doi = {10.36227/techrxiv.14528778.v1}, year = {2021}, date = {2021-05-01}, publisher = {TechRxiv}, keywords = {}, pubstate = {published}, tppubtype = {misc} } |
| 1224. | Yuan, Zhihang ; Liu, Jingze ; Li, Xingchen ; Yan, Longhao ; Chen, Haoxiang ; Wu, Bingzhe ; Yang, Yuchao ; Sun, Guangyu NAS4RRAM: neural network architecture search for inference on RRAM-based accelerators Journal Article Science China Information Sciences, 2021. @article{yuan2021, title = {NAS4RRAM: neural network architecture search for inference on RRAM-based accelerators}, author = {Yuan, Zhihang and Liu, Jingze and Li, Xingchen and Yan, Longhao and Chen, Haoxiang and Wu, Bingzhe and Yang, Yuchao and Sun, Guangyu}, url = {https://doi.org/10.1007/s11432-020-3245-7}, doi = {10.1007/s11432-020-3245-7}, year = {2021}, date = {2021-05-01}, journal = {Science China Information Sciences}, keywords = {}, pubstate = {published}, tppubtype = {article} } |
| 1223. | Zhang, Yi ; Liu, Yang ; Liu, Shirley X Neural network architecture search with AMBER Journal Article Nature Machine Intelligence, pp. 372-373, 2021. @article{Zhang2021, title = {Neural network architecture search with AMBER}, author = {Zhang, Yi and Liu, Yang and Liu, X. Shirley}, url = {https://doi.org/10.1038/s42256-021-00350-x}, year = {2021}, date = {2021-05-01}, journal = {Nature Machine Intelligence}, pages = {372-373}, keywords = {}, pubstate = {published}, tppubtype = {article} } |
| 1222. | Schoenherr, Georg P The Nonlinearity Coefficient - A Practical Guide to Neural Architecture Design PhD Thesis 2021. @phdthesis{Schoenherr2021, title = {The Nonlinearity Coefficient - A Practical Guide to Neural Architecture Design}, author = {Georg P. Schoenherr}, url = {http://reports-archive.adm.cs.cmu.edu/anon/anon/home/ftp/usr/ftp/2021/abstracts/21-110.html}, year = {2021}, date = {2021-05-01}, keywords = {}, pubstate = {published}, tppubtype = {phdthesis} } |
| 1221. | Yang, Yuxuan; Gao, Zhongke; Li, Yanli; Wang, He A CNN identified by reinforcement learning-based optimization framework for EEG-based state evaluation Journal Article Journal of Neural Engineering, 18 (4), pp. 046059, 2021. @article{Yang_2021, title = {A CNN identified by reinforcement learning-based optimization framework for EEG-based state evaluation}, author = {Yuxuan Yang and Zhongke Gao and Yanli Li and He Wang}, url = {https://doi.org/10.1088/1741-2552/abfa71}, doi = {10.1088/1741-2552/abfa71}, year = {2021}, date = {2021-05-01}, journal = {Journal of Neural Engineering}, volume = {18}, number = {4}, pages = {046059}, publisher = {IOP Publishing}, abstract = {Objective. Electroencephalogram (EEG) data, as a kind of complex time-series, is one of the most widely-used information measurements for evaluating human psychophysiological states. Recently, numerous works applied deep learning techniques, especially the convolutional neural network (CNN), into EEG-based research. The design of the hyper-parameters of the CNN model has a great influence on the performance of the model. Therefore, automatically designing these hyper-parameters can save the time and labor of experts. This leads to the appearance of the neural architecture search technique. In this paper, we propose a reinforcement learning (RL)-based step-by-step framework to efficiently search for CNN models. Approach. Specifically, the deep Q network in RL is first used to determine the depth of convolutional layers and the connection modes among layers. Then particle swarm optimization algorithm is used to fine-tune the number and size of convolution kernels. Through this step-by-step strategy, the search space can be narrowed in each step for saving the overall time cost. This framework is employed for both EEG-based sleep stage classification and driver drowsiness evaluation tasks. Main results. The results show that compared with state-of-the-art methods, the high-performance CNN models identified by the proposed optimization framework, can achieve high overall accuracy and better root mean squared error in the two tasks. Significance. Therefore, the proposed optimization framework has a great potential to provide high-performance results for other kinds of classification and prediction tasks. In this way, it can greatly save researchers’ time cost and promote broader applications of CNNs.}, keywords = {}, pubstate = {published}, tppubtype = {article} } Objective. Electroencephalogram (EEG) data, as a kind of complex time-series, is one of the most widely-used information measurements for evaluating human psychophysiological states. Recently, numerous works applied deep learning techniques, especially the convolutional neural network (CNN), into EEG-based research. The design of the hyper-parameters of the CNN model has a great influence on the performance of the model. Therefore, automatically designing these hyper-parameters can save the time and labor of experts. This leads to the appearance of the neural architecture search technique. In this paper, we propose a reinforcement learning (RL)-based step-by-step framework to efficiently search for CNN models. Approach. Specifically, the deep Q network in RL is first used to determine the depth of convolutional layers and the connection modes among layers. Then particle swarm optimization algorithm is used to fine-tune the number and size of convolution kernels. Through this step-by-step strategy, the search space can be narrowed in each step for saving the overall time cost. This framework is employed for both EEG-based sleep stage classification and driver drowsiness evaluation tasks. Main results. The results show that compared with state-of-the-art methods, the high-performance CNN models identified by the proposed optimization framework, can achieve high overall accuracy and better root mean squared error in the two tasks. Significance. Therefore, the proposed optimization framework has a great potential to provide high-performance results for other kinds of classification and prediction tasks. In this way, it can greatly save researchers’ time cost and promote broader applications of CNNs. |
| 1220. | Pan, Zheyi; Ke, Songyu; Yang, Xiaodu; Liang, Yuxuan; Yu, Yong; Zhang, Junbo; Zheng, Yu AutoSTG: Neural Architecture Search forPredictions of Spatio-Temporal Graphs Inproceedings WWW 2021, 2021. @inproceedings{PanWWW2021, title = {AutoSTG: Neural Architecture Search forPredictions of Spatio-Temporal Graphs}, author = {Zheyi Pan and Songyu Ke and Xiaodu Yang and Yuxuan Liang and Yong Yu and Junbo Zhang and Yu Zheng}, url = {http://panzheyi.cc/publication/pan2021autostg/paper.pdf}, year = {2021}, date = {2021-04-19}, booktitle = {WWW 2021}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } |
| 1219. | Kyriakides, George ; Margaritis, Konstantinos Evolving graph convolutional networks for neural architecture search Journal Article Neural Computing and Applications, 2021. @article{Kriakides2021Evolving, title = {Evolving graph convolutional networks for neural architecture search}, author = {Kyriakides, George and Margaritis, Konstantinos}, url = {https://doi.org/10.1007/s00521-021-05979-8}, doi = {10.1007/s00521-021-05979-8}, year = {2021}, date = {2021-04-15}, journal = {Neural Computing and Applications}, abstract = {As neural architecture search (NAS) becomes an increasingly adopted method to design network architectures, various methods have been proposed to speedup the process. Besides proxy evaluation tasks, weight sharing, and scaling down the evaluated architectures, performance-predicting models exhibit multiple advantages. Eliminating the need to train candidate architectures and enabling transfer learning between datasets, researchers can also utilize them as a surrogate function for Bayesian optimization. On the other hand, graph convolutional networks (GCNs) have also been increasingly adopted for various tasks, enabling deep learning techniques on graphs without feature engineering. In this paper, we employ an evolutionary-based NAS method to evolve GCNs for the problem of predicting the relative performance of various architectures included in the NAS-Bench-101 dataset. By fine-tuning the architecture generated by our methodology, we manage to achieve a Kendall’s tau correlation coefficient of 0.907 between 1050 completely unseen architectures, utilizing only 450 samples, while also outperforming a strong baseline on the same task. Furthermore, we validate our method on custom global search space architectures, generated for the Fashion-MNIST dataset.}, keywords = {}, pubstate = {published}, tppubtype = {article} } As neural architecture search (NAS) becomes an increasingly adopted method to design network architectures, various methods have been proposed to speedup the process. Besides proxy evaluation tasks, weight sharing, and scaling down the evaluated architectures, performance-predicting models exhibit multiple advantages. Eliminating the need to train candidate architectures and enabling transfer learning between datasets, researchers can also utilize them as a surrogate function for Bayesian optimization. On the other hand, graph convolutional networks (GCNs) have also been increasingly adopted for various tasks, enabling deep learning techniques on graphs without feature engineering. In this paper, we employ an evolutionary-based NAS method to evolve GCNs for the problem of predicting the relative performance of various architectures included in the NAS-Bench-101 dataset. By fine-tuning the architecture generated by our methodology, we manage to achieve a Kendall’s tau correlation coefficient of 0.907 between 1050 completely unseen architectures, utilizing only 450 samples, while also outperforming a strong baseline on the same task. Furthermore, we validate our method on custom global search space architectures, generated for the Fashion-MNIST dataset. |
| 1218. | Wang, Linnan; Xie, Saining; Li, Teng; Fonseca, Rodrigo; Tian, Yuandong Sample-Efficient Neural Architecture Search by Learning Actions for Monte Carlo Tree Search Journal Article IEEE transactions on pattern analysis and machine intelligence, PP , 2021, ISSN: 0162-8828. @article{PMID:33826511, title = {Sample-Efficient Neural Architecture Search by Learning Actions for Monte Carlo Tree Search}, author = {Linnan Wang and Saining Xie and Teng Li and Rodrigo Fonseca and Yuandong Tian}, url = {https://doi.org/10.1109/TPAMI.2021.3071343}, doi = {10.1109/tpami.2021.3071343}, issn = {0162-8828}, year = {2021}, date = {2021-04-01}, journal = {IEEE transactions on pattern analysis and machine intelligence}, volume = {PP}, abstract = {Neural Architecture Search (NAS) has emerged as a promising technique for automatic neural network design. However, existing MCTS based NAS approaches often utilize manually designed action space, which is not directly related to the performance metric to be optimized (e.g., accuracy), leading to sample-inefficient explorations of architectures. To improve the sample efficiency, this paper proposes Latent Action Neural Architecture Search (LaNAS), which learns actions to recursively partition the search space into good or bad regions that contain networks with similar performance metrics. During the search phase, as different action sequences lead to regions with different performance, the search efficiency can be significantly improved by biasing towards the good regions. On three NAS tasks, empirical results demonstrate that LaNAS is at least an order more sample efficient than baseline methods including evolutionary algorithms, Bayesian optimizations, and random search. When applied in practice, both one-shot and regular LaNAS consistently outperform existing results. Particularly, LaNAS achieves 99.0% accuracy on CIFAR-10 and 80.8% top1 accuracy at 600 MFLOPS on ImageNet in only 800 samples, significantly outperforming AmoebaNet with 33x fewer samples. Our code is publicly available at https://github.com/facebookresearch/LaMCTS.}, keywords = {}, pubstate = {published}, tppubtype = {article} } Neural Architecture Search (NAS) has emerged as a promising technique for automatic neural network design. However, existing MCTS based NAS approaches often utilize manually designed action space, which is not directly related to the performance metric to be optimized (e.g., accuracy), leading to sample-inefficient explorations of architectures. To improve the sample efficiency, this paper proposes Latent Action Neural Architecture Search (LaNAS), which learns actions to recursively partition the search space into good or bad regions that contain networks with similar performance metrics. During the search phase, as different action sequences lead to regions with different performance, the search efficiency can be significantly improved by biasing towards the good regions. On three NAS tasks, empirical results demonstrate that LaNAS is at least an order more sample efficient than baseline methods including evolutionary algorithms, Bayesian optimizations, and random search. When applied in practice, both one-shot and regular LaNAS consistently outperform existing results. Particularly, LaNAS achieves 99.0% accuracy on CIFAR-10 and 80.8% top1 accuracy at 600 MFLOPS on ImageNet in only 800 samples, significantly outperforming AmoebaNet with 33x fewer samples. Our code is publicly available at https://github.com/facebookresearch/LaMCTS. |
| 1217. | Zhang, Zhentong ; Shan, Yugang; Yuan, Jie Multi-Level Cell Progressive Differentiable Architecture Search to Improve Image Classification Accuracy Journal Article Journal of Signal Processing Systems, 2021. @article{ZhangJSPS2021, title = {Multi-Level Cell Progressive Differentiable Architecture Search to Improve Image Classification Accuracy}, author = {Zhang, Zhentong and Yugang Shan and Jie Yuan}, url = {https://doi.org/10.1007/s11265-021-01647-1}, year = {2021}, date = {2021-03-08}, journal = {Journal of Signal Processing Systems}, abstract = {In recent years, the neural architecture search has continuously made significant progress in the field of image recognition. Among them, the differentiable method has obvious advantages compared with other search methods in terms of computational cost and accuracy to deal with image classification. However, the differentiable method is usually composed of single cell, which cannot efficiently extract the features of the network. In response to this problem, we propose a multi-level cell progressive differentiable method which allows cells to have different types according to the levels of the network. In differentiable method, the gap between the search network and the evaluation one is large, and the correlation is low. We design an algorithm to improve the distribution of architecture parameters. We also optimize the loss function and use the regularization method of additional action to improve deep network performance. The method achieves good search and classification results on CIFAR10 and ImageNet (mobile setting).}, keywords = {}, pubstate = {published}, tppubtype = {article} } In recent years, the neural architecture search has continuously made significant progress in the field of image recognition. Among them, the differentiable method has obvious advantages compared with other search methods in terms of computational cost and accuracy to deal with image classification. However, the differentiable method is usually composed of single cell, which cannot efficiently extract the features of the network. In response to this problem, we propose a multi-level cell progressive differentiable method which allows cells to have different types according to the levels of the network. In differentiable method, the gap between the search network and the evaluation one is large, and the correlation is low. We design an algorithm to improve the distribution of architecture parameters. We also optimize the loss function and use the regularization method of additional action to improve deep network performance. The method achieves good search and classification results on CIFAR10 and ImageNet (mobile setting). |
| 1216. | Zimmer, Lucas; Lindauer, Marius; Hutter, Frank Auto-Pytorch: Multi-Fidelity MetaLearning for Efficient and Robust AutoDL Journal Article IEEE transactions on pattern analysis and machine intelligence, PP , 2021, ISSN: 0162-8828. @article{PMID:33750687, title = {Auto-Pytorch: Multi-Fidelity MetaLearning for Efficient and Robust AutoDL}, author = {Lucas Zimmer and Marius Lindauer and Frank Hutter}, url = {https://doi.org/10.1109/TPAMI.2021.3067763}, doi = {10.1109/tpami.2021.3067763}, issn = {0162-8828}, year = {2021}, date = {2021-03-01}, journal = {IEEE transactions on pattern analysis and machine intelligence}, volume = {PP}, abstract = {While early AutoML frameworks focused on optimizing traditional ML pipelines and their hyperparameters, a recent trend in AutoML is to focus on neural architecture search. In this paper, we introduce Auto-PyTorch, which brings the best of these two worlds together by jointly and robustly optimizing the architecture of networks and the training hyperparameters to enable fully automated deep learning (AutoDL). Auto-PyTorch achieves state-of-the-art performance on several tabular benchmarks by combining multi-fidelity optimization with portfolio construction for warmstarting and ensembling of deep neural networks (DNNs) and common baselines for tabular data. To thoroughly study our assumptions on how to design such an AutoDL system, we additionally introduce a new benchmark on learning curves for DNNs, dubbed LCBench, and run extensive ablation studies of the full Auto-PyTorch on typical AutoML benchmarks, eventually showing that Auto-PyTorch performs better than several state-of-the-art competitors on average.}, keywords = {}, pubstate = {published}, tppubtype = {article} } While early AutoML frameworks focused on optimizing traditional ML pipelines and their hyperparameters, a recent trend in AutoML is to focus on neural architecture search. In this paper, we introduce Auto-PyTorch, which brings the best of these two worlds together by jointly and robustly optimizing the architecture of networks and the training hyperparameters to enable fully automated deep learning (AutoDL). Auto-PyTorch achieves state-of-the-art performance on several tabular benchmarks by combining multi-fidelity optimization with portfolio construction for warmstarting and ensembling of deep neural networks (DNNs) and common baselines for tabular data. To thoroughly study our assumptions on how to design such an AutoDL system, we additionally introduce a new benchmark on learning curves for DNNs, dubbed LCBench, and run extensive ablation studies of the full Auto-PyTorch on typical AutoML benchmarks, eventually showing that Auto-PyTorch performs better than several state-of-the-art competitors on average. |
| 1215. | Zheng, X; Ji, R; Chen, Y; Wang, Q; Zhang, B; Ye, Q; Chen, J; Huang, F; Tian, Y MIGO-NAS: Towards Fast and Generalizable Neural Architecture Search Journal Article IEEE Transactions on Pattern Analysis & Machine Intelligence, (01), pp. 1-1, 2021, ISSN: 1939-3539. @article{9377468, title = {MIGO-NAS: Towards Fast and Generalizable Neural Architecture Search}, author = {X Zheng and R Ji and Y Chen and Q Wang and B Zhang and Q Ye and J Chen and F Huang and Y Tian}, url = {https://www.computer.org/csdl/journal/tp/5555/01/09377468/1rUNdbz4LQY}, doi = {10.1109/TPAMI.2021.3065138}, issn = {1939-3539}, year = {2021}, date = {2021-03-01}, journal = {IEEE Transactions on Pattern Analysis & Machine Intelligence}, number = {01}, pages = {1-1}, publisher = {IEEE Computer Society}, address = {Los Alamitos, CA, USA}, keywords = {}, pubstate = {published}, tppubtype = {article} } |
| 1214. | Liu, Lanlan; Zhang, Yuting; Deng, Jia; Soatto, Stefano Dynamically Grown Generative Adversarial Networks Inproceedings AAAI 2021, 2021. @inproceedings{LiuAAAI2021, title = {Dynamically Grown Generative Adversarial Networks}, author = {Lanlan Liu and Yuting Zhang and Jia Deng and Stefano Soatto}, url = {https://www.aaai.org/AAAI21Papers/AAAI-1376.LiuL.pdf}, year = {2021}, date = {2021-02-02}, booktitle = {AAAI 2021}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } |
| 1213. | Xu, Y; Xie, L; Dai, W; Zhang, X; Chen, X; Qi, G; Xiong, H; Tian, Q Partially-Connected Neural Architecture Search for Reduced Computational Redundancy Journal Article IEEE Transactions on Pattern Analysis & Machine Intelligence, (01), pp. 1-1, 2021, ISSN: 1939-3539. @article{9354953, title = {Partially-Connected Neural Architecture Search for Reduced Computational Redundancy}, author = {Y Xu and L Xie and W Dai and X Zhang and X Chen and G Qi and H Xiong and Q Tian}, url = {https://www.computer.org/csdl/journal/tp/5555/01/09354953/1rgCccYlOaQ}, doi = {10.1109/TPAMI.2021.3059510}, issn = {1939-3539}, year = {2021}, date = {2021-02-01}, journal = {IEEE Transactions on Pattern Analysis & Machine Intelligence}, number = {01}, pages = {1-1}, publisher = {IEEE Computer Society}, address = {Los Alamitos, CA, USA}, abstract = {Differentiable architecture search (DARTS) enables effective neural architecture search (NAS) using gradient descent, but suffers from high memory and computational costs. In this paper, we propose a novel approach, namely Partially-Connected DARTS (PC-DARTS), to achieve efficient and stable neural architecture search by reducing the channel and spatial redundancies of the super-network. In the channel level, partial channel connection is presented to randomly sample a small subset of channels for operation selection to accelerate the search process and suppress the over-fitting of the super-network. Side operation is introduced for bypassing (non-sampled) channels to guarantee the performance of searched architectures under extremely low sampling rates. In the spatial level, input features are down-sampled to eliminate spatial redundancy and enhance the efficiency of the mixed computation for operation selection. Furthermore, edge normalization is developed to maintain the consistency of edge selection based on channel sampling with the architectural parameters for edges. Experimental results demonstrate that the proposed approach achieves higher search speed and training stability than DARTS. PC-DARTS obtains a top-1 error rate of 2.55% on CIFAR-10 with 0.07 GPU-days for architecture search, and a state-of-the-art top-1 error rate of 24.1% on ImageNet (under the mobile setting) within 2.8 GPU-day.}, keywords = {}, pubstate = {published}, tppubtype = {article} } Differentiable architecture search (DARTS) enables effective neural architecture search (NAS) using gradient descent, but suffers from high memory and computational costs. In this paper, we propose a novel approach, namely Partially-Connected DARTS (PC-DARTS), to achieve efficient and stable neural architecture search by reducing the channel and spatial redundancies of the super-network. In the channel level, partial channel connection is presented to randomly sample a small subset of channels for operation selection to accelerate the search process and suppress the over-fitting of the super-network. Side operation is introduced for bypassing (non-sampled) channels to guarantee the performance of searched architectures under extremely low sampling rates. In the spatial level, input features are down-sampled to eliminate spatial redundancy and enhance the efficiency of the mixed computation for operation selection. Furthermore, edge normalization is developed to maintain the consistency of edge selection based on channel sampling with the architectural parameters for edges. Experimental results demonstrate that the proposed approach achieves higher search speed and training stability than DARTS. PC-DARTS obtains a top-1 error rate of 2.55% on CIFAR-10 with 0.07 GPU-days for architecture search, and a state-of-the-art top-1 error rate of 24.1% on ImageNet (under the mobile setting) within 2.8 GPU-day. |
| 1212. | Ney, Jonas; Loroch, Dominik Marek; Rybalkin, Vladimir; Weber, Nico; ü, Jens Kr; Wehn, Norbert HALF: Holistic Auto Machine Learning for FPGAs Technical Report , 2021. @techreport{DBLP:journals/corr/abs-2106-14771, title = {HALF: Holistic Auto Machine Learning for FPGAs}, author = {Jonas Ney and Dominik Marek Loroch and Vladimir Rybalkin and Nico Weber and Jens Kr ü and Norbert Wehn}, url = {https://arxiv.org/abs/2106.14771}, year = {2021}, date = {2021-01-01}, journal = {CoRR}, volume = {abs/2106.14771}, keywords = {}, pubstate = {published}, tppubtype = {techreport} } |
| 1211. | Wang, Xinyi; Xiang, Tiange; Zhang, Chaoyi; Song, Yang; Liu, Dongnan; Huang, Heng; Cai, Weidong BiX-NAS: Searching Efficient Bi-directional Architecture for Medical Image Segmentation Technical Report , 2021. @techreport{DBLP:journals/corr/abs-2106-14033, title = {BiX-NAS: Searching Efficient Bi-directional Architecture for Medical Image Segmentation}, author = {Xinyi Wang and Tiange Xiang and Chaoyi Zhang and Yang Song and Dongnan Liu and Heng Huang and Weidong Cai}, url = {https://arxiv.org/abs/2106.14033}, year = {2021}, date = {2021-01-01}, journal = {CoRR}, volume = {abs/2106.14033}, keywords = {}, pubstate = {published}, tppubtype = {techreport} } |
| 1210. | Su, Xiu; You, Shan; Xie, Jiyang; Zheng, Mingkai; Wang, Fei; Qian, Chen; Zhang, Changshui; Wang, Xiaogang; Xu, Chang Vision Transformer Architecture Search Technical Report , 2021. @techreport{DBLP:journals/corr/abs-2106-13700, title = {Vision Transformer Architecture Search}, author = {Xiu Su and Shan You and Jiyang Xie and Mingkai Zheng and Fei Wang and Chen Qian and Changshui Zhang and Xiaogang Wang and Chang Xu}, url = {https://arxiv.org/abs/2106.13700}, year = {2021}, date = {2021-01-01}, journal = {CoRR}, volume = {abs/2106.13700}, keywords = {}, pubstate = {published}, tppubtype = {techreport} } |
| 1209. | Wu, Robert; Saxena, Nayan; Jain, Rohan Poisoning the Search Space in Neural Architecture Search Technical Report , 2021. @techreport{DBLP:journals/corr/abs-2106-14406, title = {Poisoning the Search Space in Neural Architecture Search}, author = {Robert Wu and Nayan Saxena and Rohan Jain}, url = {https://arxiv.org/abs/2106.14406}, year = {2021}, date = {2021-01-01}, journal = {CoRR}, volume = {abs/2106.14406}, keywords = {}, pubstate = {published}, tppubtype = {techreport} } |
| 1208. | Negi, Shubham; Chakraborty, Indranil; Ankit, Aayush; Roy, Kaushik NAX: Co-Designing Neural Network and Hardware Architecture for Memristive Xbar based Computing Systems Technical Report , 2021. @techreport{DBLP:journals/corr/abs-2106-12125, title = {NAX: Co-Designing Neural Network and Hardware Architecture for Memristive Xbar based Computing Systems}, author = {Shubham Negi and Indranil Chakraborty and Aayush Ankit and Kaushik Roy}, url = {https://arxiv.org/abs/2106.12125}, year = {2021}, date = {2021-01-01}, journal = {CoRR}, volume = {abs/2106.12125}, keywords = {}, pubstate = {published}, tppubtype = {techreport} } |
| 1207. | Deng, Xueqing; Zhu, Yi; Tian, Yuxin; Newsam, Shawn D AutoAdapt: Automated Segmentation Network Search for Unsupervised Domain Adaptation Technical Report , 2021. @techreport{DBLP:journals/corr/abs-2106-13227, title = {AutoAdapt: Automated Segmentation Network Search for Unsupervised Domain Adaptation}, author = {Xueqing Deng and Yi Zhu and Yuxin Tian and Shawn D Newsam}, url = {https://arxiv.org/abs/2106.13227}, year = {2021}, date = {2021-01-01}, journal = {CoRR}, volume = {abs/2106.13227}, keywords = {}, pubstate = {published}, tppubtype = {techreport} } |
| 1206. | ZHAO, Huan; YAO, Quanming; TU, Weiwei Search to aggregate neighborhood for graph neural network Inproceedings 2021 IEEE 37th International Conference on Data Engineering (ICDE), pp. 552-563, 2021. @inproceedings{9458743, title = {Search to aggregate neighborhood for graph neural network}, author = {Huan ZHAO and Quanming YAO and Weiwei TU}, url = {https://ieeexplore.ieee.org/abstract/document/9458743}, doi = {10.1109/ICDE51399.2021.00054}, year = {2021}, date = {2021-01-01}, booktitle = {2021 IEEE 37th International Conference on Data Engineering (ICDE)}, pages = {552-563}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } |
| 1205. | Zhang, Miao; Su, Steven W; Pan, Shirui; Chang, Xiaojun; Abbasnejad, Ehsan M; Haffari, Reza iDARTS: Differentiable Architecture Search with Stochastic Implicit Gradients Inproceedings Meila, Marina; Zhang, Tong (Ed.): Proceedings of the 38th International Conference on Machine Learning, ICML 2021, 18-24 July 2021, Virtual Event, pp. 12557–12566, PMLR, 2021. @inproceedings{DBLP:conf/icml/ZhangSPCAH21, title = {iDARTS: Differentiable Architecture Search with Stochastic Implicit Gradients}, author = {Miao Zhang and Steven W Su and Shirui Pan and Xiaojun Chang and Ehsan M Abbasnejad and Reza Haffari}, editor = {Marina Meila and Tong Zhang}, url = {http://proceedings.mlr.press/v139/zhang21s.html}, year = {2021}, date = {2021-01-01}, booktitle = {Proceedings of the 38th International Conference on Machine Learning, ICML 2021, 18-24 July 2021, Virtual Event}, volume = {139}, pages = {12557--12566}, publisher = {PMLR}, series = {Proceedings of Machine Learning Research}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } |
| 1204. | Li, Boyang; Lu, Qing; Jiang, Weiwen; Jung, Taeho; Shi, Yiyu A mining pool solution for novel proof-of-neural-architecture consensus Inproceedings 2021 IEEE International Conference on Blockchain and Cryptocurrency (ICBC), pp. 1-3, 2021. @inproceedings{9461067, title = {A mining pool solution for novel proof-of-neural-architecture consensus}, author = {Boyang Li and Qing Lu and Weiwen Jiang and Taeho Jung and Yiyu Shi}, doi = {10.1109/ICBC51069.2021.9461067}, year = {2021}, date = {2021-01-01}, booktitle = {2021 IEEE International Conference on Blockchain and Cryptocurrency (ICBC)}, pages = {1-3}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } |
| 1203. | Nunes, Matheus; Fraga, Paulo M; Pappa, Gisele L Fitness Landscape Analysis of Graph Neural Network Architecture Search Spaces Inproceedings Proceedings of the Genetic and Evolutionary Computation Conference, pp. 876–884, Association for Computing Machinery, Lille, France, 2021, ISBN: 9781450383509. @inproceedings{10.1145/3449639.3459318, title = {Fitness Landscape Analysis of Graph Neural Network Architecture Search Spaces}, author = {Matheus Nunes and Paulo M Fraga and Gisele L Pappa}, url = {https://doi.org/10.1145/3449639.3459318}, doi = {10.1145/3449639.3459318}, isbn = {9781450383509}, year = {2021}, date = {2021-01-01}, booktitle = {Proceedings of the Genetic and Evolutionary Computation Conference}, pages = {876–884}, publisher = {Association for Computing Machinery}, address = {Lille, France}, series = {GECCO '21}, abstract = {Neural Architecture Search (NAS) is the name given to a set of methods designed to automatically configure the layout of neural networks. Their success on Convolutional Neural Networks inspired its use on optimizing other types of neural network architectures, including Graph Neural Networks (GNNs). GNNs have been extensively applied over several collections of real-world data, achieving state-of-the-art results in tasks such as circuit design, molecular structure generation and anomaly detection. Many GNN models have been recently proposed, and choosing the best model for each problem has become a cumbersome and error-prone task. Aiming to alleviate this problem, recent works have proposed strategies for applying NAS to GNN models. However, different search methods converge relatively fast in the search for a good architecture, which raises questions about the structure of the problem. In this work we use Fitness Landscape Analysis (FLA) measures to characterize the search space explored by NAS methods for GNNs. We sample almost 90k different architectures that cover most of the fitness range, and represent them using both a one-hot encoding and an embedding representation. Results of the fitness distance correlation and dispersion metrics show the fitness landscape is easy to be explored, and presents low neutrality.}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } Neural Architecture Search (NAS) is the name given to a set of methods designed to automatically configure the layout of neural networks. Their success on Convolutional Neural Networks inspired its use on optimizing other types of neural network architectures, including Graph Neural Networks (GNNs). GNNs have been extensively applied over several collections of real-world data, achieving state-of-the-art results in tasks such as circuit design, molecular structure generation and anomaly detection. Many GNN models have been recently proposed, and choosing the best model for each problem has become a cumbersome and error-prone task. Aiming to alleviate this problem, recent works have proposed strategies for applying NAS to GNN models. However, different search methods converge relatively fast in the search for a good architecture, which raises questions about the structure of the problem. In this work we use Fitness Landscape Analysis (FLA) measures to characterize the search space explored by NAS methods for GNNs. We sample almost 90k different architectures that cover most of the fitness range, and represent them using both a one-hot encoding and an embedding representation. Results of the fitness distance correlation and dispersion metrics show the fitness landscape is easy to be explored, and presents low neutrality. |
| 1202. | Zhang, Miao; Su, Steven W; Pan, Shirui; Chang, Xiaojun; Huang, Wei; Haffari, Gholamreza Differentiable Architecture Search Without Training Nor Labels: A Pruning Perspective Technical Report , 2021. @techreport{DBLP:journals/corr/abs-2106-11542, title = {Differentiable Architecture Search Without Training Nor Labels: A Pruning Perspective}, author = {Miao Zhang and Steven W Su and Shirui Pan and Xiaojun Chang and Wei Huang and Gholamreza Haffari}, url = {https://arxiv.org/abs/2106.11542}, year = {2021}, date = {2021-01-01}, journal = {CoRR}, volume = {abs/2106.11542}, keywords = {}, pubstate = {published}, tppubtype = {techreport} } |
| 1201. | Klos, Andreas; Rosenbaum, Marius; Schiffmann, Wolfram Scalable and Highly Available Multi-Objective Neural Architecture Search in Bare Metal Kubernetes Cluster Inproceedings 2021 IEEE International Parallel and Distributed Processing Symposium Workshops (IPDPSW), pp. 605-610, 2021. @inproceedings{9460405, title = {Scalable and Highly Available Multi-Objective Neural Architecture Search in Bare Metal Kubernetes Cluster}, author = {Andreas Klos and Marius Rosenbaum and Wolfram Schiffmann}, url = {https://ieeexplore.ieee.org/abstract/document/9460405}, doi = {10.1109/IPDPSW52791.2021.00094}, year = {2021}, date = {2021-01-01}, booktitle = {2021 IEEE International Parallel and Distributed Processing Symposium Workshops (IPDPSW)}, pages = {605-610}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } |
| 1200. | Eriksson, David; -, Pierce I; Daulton, Samuel; Xia, Peng; Shrivastava, Akshat; Babu, Arun; Zhao, Shicong; Aly, Ahmed; Venkatesh, Ganesh; Balandat, Maximilian Latency-Aware Neural Architecture Search with Multi-Objective Bayesian Optimization Inproceedings 8th ICML Workshop on Automated Machine Learning (2021), 2021. @inproceedings{DBLP:journals/corr/abs-2106-11890, title = {Latency-Aware Neural Architecture Search with Multi-Objective Bayesian Optimization}, author = {David Eriksson and Pierce I - and Samuel Daulton and Peng Xia and Akshat Shrivastava and Arun Babu and Shicong Zhao and Ahmed Aly and Ganesh Venkatesh and Maximilian Balandat}, url = {https://arxiv.org/abs/2106.11890}, year = {2021}, date = {2021-01-01}, booktitle = {8th ICML Workshop on Automated Machine Learning (2021)}, journal = {CoRR}, volume = {abs/2106.11890}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } |
| 1199. | Zeinali, Behnam; Zhuang, Di; Chang, Morris J ESAI: Efficient Split Artificial Intelligence via Early Exiting Using Neural Architecture Search Technical Report , 2021. @techreport{DBLP:journals/corr/abs-2106-12549, title = {ESAI: Efficient Split Artificial Intelligence via Early Exiting Using Neural Architecture Search}, author = {Behnam Zeinali and Di Zhuang and Morris J Chang}, url = {https://arxiv.org/abs/2106.12549}, year = {2021}, date = {2021-01-01}, journal = {CoRR}, volume = {abs/2106.12549}, keywords = {}, pubstate = {published}, tppubtype = {techreport} } |
| 1198. | Santra, Santanu; Hsieh, Jun-Wei; Lin, Chi-Fang Gradient Descent Effects on Differential Neural Architecture Search: A Survey Journal Article IEEE Access, 9 , pp. 89602-89618, 2021. @article{9461192, title = {Gradient Descent Effects on Differential Neural Architecture Search: A Survey}, author = {Santanu Santra and Jun-Wei Hsieh and Chi-Fang Lin}, url = {https://ieeexplore.ieee.org/abstract/document/9461192}, doi = {10.1109/ACCESS.2021.3090918}, year = {2021}, date = {2021-01-01}, journal = {IEEE Access}, volume = {9}, pages = {89602-89618}, keywords = {}, pubstate = {published}, tppubtype = {article} } |
| 1197. | Maile, Kaitlin; Lecarpentier, Erwan; é, Herv; Wilson, Dennis G On Constrained Optimization in Differentiable Neural Architecture Search Technical Report , 2021. @techreport{DBLP:journals/corr/abs-2106-11655, title = {On Constrained Optimization in Differentiable Neural Architecture Search}, author = {Kaitlin Maile and Erwan Lecarpentier and Herv é and Dennis G Wilson}, url = {https://arxiv.org/abs/2106.11655}, year = {2021}, date = {2021-01-01}, journal = {CoRR}, volume = {abs/2106.11655}, keywords = {}, pubstate = {published}, tppubtype = {techreport} } |
| 1196. | Azarmehr, Neda; Ye, Xujiong; Howard, James P; Lane, Elisabeth S; Labs, Robert; Shun-Shin, Matthew J; Cole, Graham D; Bidaut, Luc; Francis, Darrel P; Zolgharni, Massoud Neural architecture search of echocardiography view classifiers Journal Article Journal of Medical Imaging, 8 (3), pp. 1 – 21, 2021. @article{10.1117/1.JMI.8.3.034002, title = {Neural architecture search of echocardiography view classifiers}, author = {Neda Azarmehr and Xujiong Ye and James P Howard and Elisabeth S Lane and Robert Labs and Matthew J Shun-Shin and Graham D Cole and Luc Bidaut and Darrel P Francis and Massoud Zolgharni}, url = {https://doi.org/10.1117/1.JMI.8.3.034002}, doi = {10.1117/1.JMI.8.3.034002}, year = {2021}, date = {2021-01-01}, journal = {Journal of Medical Imaging}, volume = {8}, number = {3}, pages = {1 -- 21}, publisher = {SPIE}, keywords = {}, pubstate = {published}, tppubtype = {article} } |
| 1195. | Li, Yuening; Chen, Zhengzhang; Zha, Daochen; Zhou, Kaixiong; Jin, Haifeng; Chen, Haifeng; Hu, Xia AutoOD: Neural Architecture Search for Outlier Detection Inproceedings 2021 IEEE 37th International Conference on Data Engineering (ICDE), pp. 2117-2122, 2021. @inproceedings{9458691, title = {AutoOD: Neural Architecture Search for Outlier Detection}, author = {Yuening Li and Zhengzhang Chen and Daochen Zha and Kaixiong Zhou and Haifeng Jin and Haifeng Chen and Xia Hu}, url = {https://ieeexplore.ieee.org/abstract/document/9458691}, doi = {10.1109/ICDE51399.2021.00210}, year = {2021}, date = {2021-01-01}, booktitle = {2021 IEEE 37th International Conference on Data Engineering (ICDE)}, pages = {2117-2122}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } |
| 1194. | Manu, Daniel; Huang, Shaoyi; Ding, Caiwen; Yang, Lei Co-Exploration of Graph Neural Network and Network-on-Chip Design Using AutoML Inproceedings Proceedings of the 2021 on Great Lakes Symposium on VLSI, pp. 175–180, Association for Computing Machinery, Virtual Event, USA, 2021, ISBN: 9781450383936. @inproceedings{10.1145/3453688.3461741, title = {Co-Exploration of Graph Neural Network and Network-on-Chip Design Using AutoML}, author = {Daniel Manu and Shaoyi Huang and Caiwen Ding and Lei Yang}, url = {https://doi.org/10.1145/3453688.3461741}, doi = {10.1145/3453688.3461741}, isbn = {9781450383936}, year = {2021}, date = {2021-01-01}, booktitle = {Proceedings of the 2021 on Great Lakes Symposium on VLSI}, pages = {175–180}, publisher = {Association for Computing Machinery}, address = {Virtual Event, USA}, series = {GLSVLSI '21}, abstract = {Recently, Graph Neural Networks (GNNs) have exhibited high efficiency in several graph-based machine learning tasks. Compared with the neural networks for computer vision or speech tasks (e.g., Convolutional Neural Networks), GNNs have much higher requirements on communication due to the complicated graph structures; however, when applying GNNs for real-world applications, say in recommender systems (e.g. Uber Eats), it commonly has the real-time requirements. To deal with the tradeoff between the complicated architecture and the high-demand timing performance, both GNN architecture and hardware accelerator need to be optimized. Network-on-Chip (NoC), derived for efficiently managing the high-volume of communications, naturally becomes one of the top candidates to accelerate GNNs. However, there is a missing link between the optimize of GNN architecture and the NoC design.In this work, we present an AutoML-based framework GN-NAS, aiming at searching for the optimum GNN architecture, which can be suitable for the NoC accelerator. We devise a robust reinforcement learning based controller to validate the retained best GNN architectures, coupled with a parameter sharing approach, namely ParamShare, to improve search efficiency. Experimental results on four graph-based benchmark datasets, Cora, Citeseer, Pubmed and Protein-Protein Interaction show that the GNN architectures obtained by our framework outperform that of the state-of-the-art and baseline models, whilst reducing model size which makes them easy to deploy onto the NoC platform.}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } Recently, Graph Neural Networks (GNNs) have exhibited high efficiency in several graph-based machine learning tasks. Compared with the neural networks for computer vision or speech tasks (e.g., Convolutional Neural Networks), GNNs have much higher requirements on communication due to the complicated graph structures; however, when applying GNNs for real-world applications, say in recommender systems (e.g. Uber Eats), it commonly has the real-time requirements. To deal with the tradeoff between the complicated architecture and the high-demand timing performance, both GNN architecture and hardware accelerator need to be optimized. Network-on-Chip (NoC), derived for efficiently managing the high-volume of communications, naturally becomes one of the top candidates to accelerate GNNs. However, there is a missing link between the optimize of GNN architecture and the NoC design.In this work, we present an AutoML-based framework GN-NAS, aiming at searching for the optimum GNN architecture, which can be suitable for the NoC accelerator. We devise a robust reinforcement learning based controller to validate the retained best GNN architectures, coupled with a parameter sharing approach, namely ParamShare, to improve search efficiency. Experimental results on four graph-based benchmark datasets, Cora, Citeseer, Pubmed and Protein-Protein Interaction show that the GNN architectures obtained by our framework outperform that of the state-of-the-art and baseline models, whilst reducing model size which makes them easy to deploy onto the NoC platform. |
| 1193. | Chen, Weiwei; Wang, Ying; Lin, Gangliang; Gao, Chengsi; Liu, Cheng; Zhang, Lei CHaNAS: Coordinated Search for Network Architecture and Scheduling Policy Inproceedings Proceedings of the 22nd ACM SIGPLAN/SIGBED International Conference on Languages, Compilers, and Tools for Embedded Systems, pp. 42–53, Association for Computing Machinery, Virtual, Canada, 2021, ISBN: 9781450384728. @inproceedings{10.1145/3461648.3463846, title = {CHaNAS: Coordinated Search for Network Architecture and Scheduling Policy}, author = {Weiwei Chen and Ying Wang and Gangliang Lin and Chengsi Gao and Cheng Liu and Lei Zhang}, url = {https://doi.org/10.1145/3461648.3463846}, doi = {10.1145/3461648.3463846}, isbn = {9781450384728}, year = {2021}, date = {2021-01-01}, booktitle = {Proceedings of the 22nd ACM SIGPLAN/SIGBED International Conference on Languages, Compilers, and Tools for Embedded Systems}, pages = {42–53}, publisher = {Association for Computing Machinery}, address = {Virtual, Canada}, series = {LCTES 2021}, abstract = {Automatically design an efficient DNN solution for a given deep learning task on the target hardware mainly decided by the neural network architecture and the schedule mapping strategy, where the two goals are closely coupled with each other to fully exploit the advantages of the underlying hardware. Prior hardware-aware Neural Architecture Search (NAS) methods mostly ignore the impacts of different scheduling policies (e.g., graph-level optimization, loop transformations, parallelization, etc.) on network candidates being evaluated in the search process. Thus, they may miss the true-optimal architecture that can only be discovered by trying-out different scheduling policies. This work proposes a NAS framework (CHaNAS) that searches for not only the network architecture but also the dedicated scheduling policy, as the optimal co-design solution on target hardware that fully exploits the advantages of the underlying hardware. We propose to use a block-based pre-scheduling methodology to reduce the co-design search space, and enable the automatic generation of the optimal co-design, including the network architecture and the tensor programs that practice the scheduling policy. We evaluate CHaNAS on Imagenet on different hardware back-ends against the state-of-the-art hardware-aware search method MobileNet-v3. Experimental results show that the co-design solutions obtained by ChaNAS show up to 1.6x, 1.9x, and 1.7x performance boost on NVIDIA P100 GPU, Intel Xeon 8163 CPU, and Samsung Note 10 Mobile, respectively, over the baselines of the same-level accuracy.}, keywords = {}, pubstate = {published}, tppubtype = {inproceedings} } Automatically design an efficient DNN solution for a given deep learning task on the target hardware mainly decided by the neural network architecture and the schedule mapping strategy, where the two goals are closely coupled with each other to fully exploit the advantages of the underlying hardware. Prior hardware-aware Neural Architecture Search (NAS) methods mostly ignore the impacts of different scheduling policies (e.g., graph-level optimization, loop transformations, parallelization, etc.) on network candidates being evaluated in the search process. Thus, they may miss the true-optimal architecture that can only be discovered by trying-out different scheduling policies. This work proposes a NAS framework (CHaNAS) that searches for not only the network architecture but also the dedicated scheduling policy, as the optimal co-design solution on target hardware that fully exploits the advantages of the underlying hardware. We propose to use a block-based pre-scheduling methodology to reduce the co-design search space, and enable the automatic generation of the optimal co-design, including the network architecture and the tensor programs that practice the scheduling policy. We evaluate CHaNAS on Imagenet on different hardware back-ends against the state-of-the-art hardware-aware search method MobileNet-v3. Experimental results show that the co-design solutions obtained by ChaNAS show up to 1.6x, 1.9x, and 1.7x performance boost on NVIDIA P100 GPU, Intel Xeon 8163 CPU, and Samsung Note 10 Mobile, respectively, over the baselines of the same-level accuracy. |
| 1192. | Lee, Hayeon; Lee, Sewoong; Chong, Song; Hwang, Sung Ju HELP: Hardware-Adaptive Efficient Latency Predictor for NAS via Meta-Learning Technical Report , 2021. @techreport{DBLP:journals/corr/abs-2106-08630, title = {HELP: Hardware-Adaptive Efficient Latency Predictor for NAS via Meta-Learning}, author = {Hayeon Lee and Sewoong Lee and Song Chong and Sung Ju Hwang}, url = {https://arxiv.org/abs/2106.08630}, year = {2021}, date = {2021-01-01}, journal = {CoRR}, volume = {abs/2106.08630}, keywords = {}, pubstate = {published}, tppubtype = {techreport} } |
| 1191. | Li, Hanjun; Wu, Gaojie; -, Wei Combined Depth Space based Architecture Search For Person Re-identification Incollection CVPR2021, abs/2104.04163 , 2021. @incollection{DBLP:journals/corr/abs-2104-04163, title = {Combined Depth Space based Architecture Search For Person Re-identification}, author = {Hanjun Li and Gaojie Wu and Wei -}, url = {https://arxiv.org/abs/2104.04163}, year = {2021}, date = {2021-01-01}, booktitle = {CVPR2021}, journal = {CoRR}, volume = {abs/2104.04163}, keywords = {}, pubstate = {published}, tppubtype = {incollection} } |
| 1190. | Garciarena, Unai; Santana, Roberto; Mendiburu, Alexander Redefining Neural Architecture Search of Heterogeneous Multi-Network Models by Characterizing Variation Operators and Model Components Technical Report , 2021. @techreport{DBLP:journals/corr/abs-2106-08972, title = {Redefining Neural Architecture Search of Heterogeneous Multi-Network Models by Characterizing Variation Operators and Model Components}, author = {Unai Garciarena and Roberto Santana and Alexander Mendiburu}, url = {https://arxiv.org/abs/2106.08972}, year = {2021}, date = {2021-01-01}, journal = {CoRR}, volume = {abs/2106.08972}, keywords = {}, pubstate = {published}, tppubtype = {techreport} } |
| 1189. | Akhauri, Yash; Niranjan, Adithya; ñ, Pablo Mu J; Banerjee, Suvadeep; Davare, Abhijit; Cocchini, Pasquale; Sorokin, Anton A; Iyer, Ravi; Jain, Nilesh RHNAS: Realizable Hardware and Neural Architecture Search Technical Report , 2021. @techreport{DBLP:journals/corr/abs-2106-09180, title = {RHNAS: Realizable Hardware and Neural Architecture Search}, author = {Yash Akhauri and Adithya Niranjan and Pablo Mu J ñ and Suvadeep Banerjee and Abhijit Davare and Pasquale Cocchini and Anton A Sorokin and Ravi Iyer and Nilesh Jain}, url = {https://arxiv.org/abs/2106.09180}, year = {2021}, date = {2021-01-01}, journal = {CoRR}, volume = {abs/2106.09180}, keywords = {}, pubstate = {published}, tppubtype = {techreport} } |
| 1188. | Fu, Yonggan; Zhang, Yongan; Zhang, Yang; Cox, David; Lin, Yingyan Auto-NBA: Efficient and Effective Search Over the Joint Space of Networks, Bitwidths, and Accelerators Technical Report , 2021. @techreport{DBLP:journals/corr/abs-2106-06575, title = {Auto-NBA: Efficient and Effective Search Over the Joint Space of Networks, Bitwidths, and Accelerators}, author = {Yonggan Fu and Yongan Zhang and Yang Zhang and David Cox and Yingyan Lin}, url = {https://arxiv.org/abs/2106.06575}, year = {2021}, date = {2021-01-01}, journal = {CoRR}, volume = {abs/2106.06575}, keywords = {}, pubstate = {published}, tppubtype = {techreport} } |