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.
Transformers have gained increasing popularity in different domains. For a comprehensive list of papers focusing on Neural Architecture Search for Transformer-Based spaces, the awesome-transformer-search repo is all you need.
2024
3217.
Liang, Xinyan; Fu, Pinhan; Guo, Qian; Zheng, Keyin; Qian, Yuhua
DC-NAS: Divide-and-Conquer Neural Architecture Search for Multi-Modal Classification Proceedings Article
In: Wooldridge, Michael J.; Dy, Jennifer G.; Natarajan, Sriraam (Ed.): Thirty-Eighth AAAI Conference on Artificial Intelligence, AAAI 2024, Thirty-Sixth Conference on Innovative Applications of Artificial Intelligence, IAAI 2024, Fourteenth Symposium on Educational Advances in Artificial Intelligence, EAAI 2014, February 20-27, 2024, Vancouver, Canada, pp. 13754–13762, AAAI Press, 2024.
@inproceedings{DBLP:conf/aaai/LiangFGZQ24,
title = {DC-NAS: Divide-and-Conquer Neural Architecture Search for Multi-Modal Classification},
author = {Xinyan Liang and Pinhan Fu and Qian Guo and Keyin Zheng and Yuhua Qian},
editor = {Michael J. Wooldridge and Jennifer G. Dy and Sriraam Natarajan},
url = {https://doi.org/10.1609/aaai.v38i12.29281},
doi = {10.1609/AAAI.V38I12.29281},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
booktitle = {Thirty-Eighth AAAI Conference on Artificial Intelligence, AAAI
2024, Thirty-Sixth Conference on Innovative Applications of Artificial
Intelligence, IAAI 2024, Fourteenth Symposium on Educational Advances
in Artificial Intelligence, EAAI 2014, February 20-27, 2024, Vancouver,
Canada},
pages = {13754–13762},
publisher = {AAAI Press},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
3216.
Lin, Wei; Peng, Xu; Yu, Zhengtao; Jin, Taisong
Hypergraph Neural Architecture Search Proceedings Article
In: Wooldridge, Michael J.; Dy, Jennifer G.; Natarajan, Sriraam (Ed.): Thirty-Eighth AAAI Conference on Artificial Intelligence, AAAI 2024, Thirty-Sixth Conference on Innovative Applications of Artificial Intelligence, IAAI 2024, Fourteenth Symposium on Educational Advances in Artificial Intelligence, EAAI 2014, February 20-27, 2024, Vancouver, Canada, pp. 13837–13845, AAAI Press, 2024.
@inproceedings{DBLP:conf/aaai/LinP0J24,
title = {Hypergraph Neural Architecture Search},
author = {Wei Lin and Xu Peng and Zhengtao Yu and Taisong Jin},
editor = {Michael J. Wooldridge and Jennifer G. Dy and Sriraam Natarajan},
url = {https://doi.org/10.1609/aaai.v38i12.29290},
doi = {10.1609/AAAI.V38I12.29290},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
booktitle = {Thirty-Eighth AAAI Conference on Artificial Intelligence, AAAI
2024, Thirty-Sixth Conference on Innovative Applications of Artificial
Intelligence, IAAI 2024, Fourteenth Symposium on Educational Advances
in Artificial Intelligence, EAAI 2014, February 20-27, 2024, Vancouver,
Canada},
pages = {13837–13845},
publisher = {AAAI Press},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
3215.
Wu, Yongtao; Liu, Fanghui; Simon-Gabriel, Carl-Johann; Chrysos, Grigorios G.; Cevher, Volkan
Robust NAS under adversarial training: benchmark, theory, and beyond Technical Report
2024.
@techreport{DBLP:journals/corr/abs-2403-13134,
title = {Robust NAS under adversarial training: benchmark, theory, and beyond},
author = {Yongtao Wu and Fanghui Liu and Carl-Johann Simon-Gabriel and Grigorios G. Chrysos and Volkan Cevher},
url = {https://doi.org/10.48550/arXiv.2403.13134},
doi = {10.48550/ARXIV.2403.13134},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {CoRR},
volume = {abs/2403.13134},
keywords = {},
pubstate = {published},
tppubtype = {techreport}
}
3214.
Mills, Keith G.; Han, Fred X.; Salameh, Mohammad; Lu, Shengyao; Zhou, Chunhua; He, Jiao; Sun, Fengyu; Niu, Di
Building Optimal Neural Architectures using Interpretable Knowledge Technical Report
2024.
@techreport{DBLP:journals/corr/abs-2403-13293,
title = {Building Optimal Neural Architectures using Interpretable Knowledge},
author = {Keith G. Mills and Fred X. Han and Mohammad Salameh and Shengyao Lu and Chunhua Zhou and Jiao He and Fengyu Sun and Di Niu},
url = {https://doi.org/10.48550/arXiv.2403.13293},
doi = {10.48550/ARXIV.2403.13293},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {CoRR},
volume = {abs/2403.13293},
keywords = {},
pubstate = {published},
tppubtype = {techreport}
}
3213.
Rajapakshe, Thejan; Rana, Rajib; Khalifa, Sara; Sisman, Berrak; Schuller, Björn W.; Busso, Carlos
emoDARTS: Joint Optimisation of CNN & Sequential Neural Network Architectures for Superior Speech Emotion Recognition Technical Report
2024.
@techreport{DBLP:journals/corr/abs-2403-14083,
title = {emoDARTS: Joint Optimisation of CNN & Sequential Neural Network Architectures for Superior Speech Emotion Recognition},
author = {Thejan Rajapakshe and Rajib Rana and Sara Khalifa and Berrak Sisman and Björn W. Schuller and Carlos Busso},
url = {https://doi.org/10.48550/arXiv.2403.14083},
doi = {10.48550/ARXIV.2403.14083},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {CoRR},
volume = {abs/2403.14083},
keywords = {},
pubstate = {published},
tppubtype = {techreport}
}
3212.
Benmeziane, Hadjer; Maghraoui, Kaoutar El; Ouarnoughi, Hamza; Niar, Smail
Grassroots operator search for model edge adaptation using mathematical search space Journal Article
In: Future Generation Computer Systems, vol. 157, pp. 29-40, 2024, ISSN: 0167-739X.
@article{BENMEZIANE202429,
title = {Grassroots operator search for model edge adaptation using mathematical search space},
author = {Hadjer Benmeziane and Kaoutar El Maghraoui and Hamza Ouarnoughi and Smail Niar},
url = {https://www.sciencedirect.com/science/article/pii/S0167739X24001018},
doi = {https://doi.org/10.1016/j.future.2024.03.029},
issn = {0167-739X},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {Future Generation Computer Systems},
volume = {157},
pages = {29-40},
abstract = {Hardware-aware Neural Architecture Search (HW-NAS) is increasingly being used to design efficient deep learning architectures. An efficient and flexible search space is crucial to the success of HW-NAS. Current approaches focus on designing a macro-architecture and searching for the architecture’s hyperparameters based on a set of possible values. This approach is biased by the expertise of deep learning (DL) engineers and standard modeling approaches. In this paper, we present a Grassroots Operator Search (GOS) methodology. Our HW-NAS adapts a given model for edge devices by searching for efficient operator replacement. We express each operator as a set of mathematical instructions that capture its behavior. The mathematical instructions are then used as the basis for searching and selecting efficient replacement operators that maintain the accuracy of the original model while reducing computational complexity. Our approach is grassroots since it relies on the mathematical foundations to construct new and efficient operators for DL architectures. We demonstrate on various DL models, that our method consistently outperforms the original models on two edge devices, namely Redmi Note 7S and Raspberry Pi3, with a minimum of 2.2x speedup while maintaining high accuracy. Additionally, we showcase a use case of our GOS approach in pulse rate estimation on wristband devices, where we achieve state-of-the-art performance, while maintaining reduced computational complexity, demonstrating the effectiveness of our approach in practical applications.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Hardware-aware Neural Architecture Search (HW-NAS) is increasingly being used to design efficient deep learning architectures. An efficient and flexible search space is crucial to the success of HW-NAS. Current approaches focus on designing a macro-architecture and searching for the architecture’s hyperparameters based on a set of possible values. This approach is biased by the expertise of deep learning (DL) engineers and standard modeling approaches. In this paper, we present a Grassroots Operator Search (GOS) methodology. Our HW-NAS adapts a given model for edge devices by searching for efficient operator replacement. We express each operator as a set of mathematical instructions that capture its behavior. The mathematical instructions are then used as the basis for searching and selecting efficient replacement operators that maintain the accuracy of the original model while reducing computational complexity. Our approach is grassroots since it relies on the mathematical foundations to construct new and efficient operators for DL architectures. We demonstrate on various DL models, that our method consistently outperforms the original models on two edge devices, namely Redmi Note 7S and Raspberry Pi3, with a minimum of 2.2x speedup while maintaining high accuracy. Additionally, we showcase a use case of our GOS approach in pulse rate estimation on wristband devices, where we achieve state-of-the-art performance, while maintaining reduced computational complexity, demonstrating the effectiveness of our approach in practical applications.
3211.
Yu, Hongyuan; Wan, Cheng; Liu, Mengchen; Chen, Dongdong; Xiao, Bin; Dai, Xiyang
Real-Time Image Segmentation via Hybrid Convolutional-Transformer Architecture Search Technical Report
2024.
@techreport{DBLP:journals/corr/abs-2403-10413,
title = {Real-Time Image Segmentation via Hybrid Convolutional-Transformer Architecture Search},
author = {Hongyuan Yu and Cheng Wan and Mengchen Liu and Dongdong Chen and Bin Xiao and Xiyang Dai},
url = {https://doi.org/10.48550/arXiv.2403.10413},
doi = {10.48550/ARXIV.2403.10413},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {CoRR},
volume = {abs/2403.10413},
keywords = {},
pubstate = {published},
tppubtype = {techreport}
}
3210.
Zhou, Qinqin; Sheng, Kekai; Zheng, Xiawu; Li, Ke; Tian, Yonghong; Chen, Jie; Ji, Rongrong
Training-Free Transformer Architecture Search With Zero-Cost Proxy Guided Evolution Journal Article
In: IEEE Transactions on Pattern Analysis and Machine Intelligence, pp. 1-17, 2024.
@article{10475573,
title = {Training-Free Transformer Architecture Search With Zero-Cost Proxy Guided Evolution},
author = {Qinqin Zhou and Kekai Sheng and Xiawu Zheng and Ke Li and Yonghong Tian and Jie Chen and Rongrong Ji},
url = {https://ieeexplore.ieee.org/document/10475573},
doi = {10.1109/TPAMI.2024.3378781},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
pages = {1-17},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
3209.
Kancharagunta, Kishan Babu; Marapatla, AjayDilipKumar; Manchala, Harikesh; Pillalamarri, Uday Kiran; Kanthala, Danush; Musku, Sai Ritish Reddy; Kamsani, Charan Sai
Cyclone Intensity Estimation: A Studyof Neural Architecture Search andTransfer Learning in CNNs Miscellaneous
2024.
@misc{PPR:PPR823571,
title = {Cyclone Intensity Estimation: A Studyof Neural Architecture Search andTransfer Learning in CNNs},
author = {Kishan Babu Kancharagunta and AjayDilipKumar Marapatla and Harikesh Manchala and Uday Kiran Pillalamarri and Danush Kanthala and Sai Ritish Reddy Musku and Charan Sai Kamsani},
url = {https://doi.org/10.21203/rs.3.rs-4093387/v1},
doi = {10.21203/rs.3.rs-4093387/v1},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
publisher = {Research Square},
abstract = {Tropical cyclones (TCs) are severe weather phenomena that can significantly affect human lives. These events can lead to calamities characterized by strong sustainable winds and enormous waves. We proposed an architecture based on convolutional neural networks (CNNs) to tackle this problem. This method makes use of cyclone infrared images. Using customized FCL architectures, we used transfer learning and fine-tuning on CNN architectures such as VGG16, VGG19, and ResNet50, both with and without data augmentation. Fine-tuning involved 4 layers of VGG16, 8 layers of VGG19, and 12 layers of ResNet50 to capture cyclone features effectively. The CNN models were used to these architectures to extract features, and the resulting feature maps were fed to various combinations of Fully Connected Networks (FCL). The most optimistic results were achieved with the VGG16 + FCL (128 x 64 x 1) architecture through transfer learning, producing a Mean Absolute Error (MAE) of 7.51 kts, Root Mean Square Error (RMSE) of 9.63 kts, and an R2 Score of 0.92. Consequently, we identified this model as the foundational basis for Neural Architecture Search (NAS) to enhance the FCL architecture. The NAS process generated various architectures, among which the VGG16 + FCL (128 x 128 x 1) architecture stood out with notable performance, featuring a Mean Squared Error (MSE) of 6.77 kts, RMSE of 8.88 kts, and an impressive R2-Score of 0.945.},
keywords = {},
pubstate = {published},
tppubtype = {misc}
}
Tropical cyclones (TCs) are severe weather phenomena that can significantly affect human lives. These events can lead to calamities characterized by strong sustainable winds and enormous waves. We proposed an architecture based on convolutional neural networks (CNNs) to tackle this problem. This method makes use of cyclone infrared images. Using customized FCL architectures, we used transfer learning and fine-tuning on CNN architectures such as VGG16, VGG19, and ResNet50, both with and without data augmentation. Fine-tuning involved 4 layers of VGG16, 8 layers of VGG19, and 12 layers of ResNet50 to capture cyclone features effectively. The CNN models were used to these architectures to extract features, and the resulting feature maps were fed to various combinations of Fully Connected Networks (FCL). The most optimistic results were achieved with the VGG16 + FCL (128 x 64 x 1) architecture through transfer learning, producing a Mean Absolute Error (MAE) of 7.51 kts, Root Mean Square Error (RMSE) of 9.63 kts, and an R2 Score of 0.92. Consequently, we identified this model as the foundational basis for Neural Architecture Search (NAS) to enhance the FCL architecture. The NAS process generated various architectures, among which the VGG16 + FCL (128 x 128 x 1) architecture stood out with notable performance, featuring a Mean Squared Error (MSE) of 6.77 kts, RMSE of 8.88 kts, and an impressive R2-Score of 0.945.
3208.
Zhang, Beichen; Wang, Xiaoxing; Qin, Xiaohan; Yan, Junchi
Boosting Order-Preserving and Transferability for Neural Architecture Search: a Joint Architecture Refined Search and Fine-tuning Approach Technical Report
2024.
@techreport{DBLP:journals/corr/abs-2403-11380,
title = {Boosting Order-Preserving and Transferability for Neural Architecture Search: a Joint Architecture Refined Search and Fine-tuning Approach},
author = {Beichen Zhang and Xiaoxing Wang and Xiaohan Qin and Junchi Yan},
url = {https://doi.org/10.48550/arXiv.2403.11380},
doi = {10.48550/ARXIV.2403.11380},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {CoRR},
volume = {abs/2403.11380},
keywords = {},
pubstate = {published},
tppubtype = {techreport}
}
3207.
Zhong, Rui; Xu, Yuefeng; Zhang, Chao; Yu, Jun
Efficient Multiplayer Battle Game Optimizer for Adversarial Robust Neural Architecture Search Bachelor Thesis
2024.
@bachelorthesis{DBLP:journals/corr/abs-2403-10100,
title = {Efficient Multiplayer Battle Game Optimizer for Adversarial Robust Neural Architecture Search},
author = {Rui Zhong and Yuefeng Xu and Chao Zhang and Jun Yu},
url = {https://doi.org/10.48550/arXiv.2403.10100},
doi = {10.48550/ARXIV.2403.10100},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {CoRR},
volume = {abs/2403.10100},
keywords = {},
pubstate = {published},
tppubtype = {bachelorthesis}
}
3206.
Sharifi, Ali Asghar; Zoljodi, Ali; Daneshtalab, Masoud
TrajectoryNAS: A Neural Architecture Search for Trajectory Prediction Technical Report
2024.
@techreport{DBLP:journals/corr/abs-2403-11695,
title = {TrajectoryNAS: A Neural Architecture Search for Trajectory Prediction},
author = {Ali Asghar Sharifi and Ali Zoljodi and Masoud Daneshtalab},
url = {https://doi.org/10.48550/arXiv.2403.11695},
doi = {10.48550/ARXIV.2403.11695},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {CoRR},
volume = {abs/2403.11695},
keywords = {},
pubstate = {published},
tppubtype = {techreport}
}
3205.
Chen, Guanghui; Wang, Zheng; Jia, Yi; Huang, Yongming; Yang, Luxi
An Efficient Architecture Search for Scalable Beamforming Design in Cell-Free Systems Journal Article
In: IEEE Transactions on Vehicular Technology, pp. 1-13, 2024.
@article{10466649,
title = {An Efficient Architecture Search for Scalable Beamforming Design in Cell-Free Systems},
author = {Guanghui Chen and Zheng Wang and Yi Jia and Yongming Huang and Luxi Yang},
doi = {10.1109/TVT.2024.3369748},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {IEEE Transactions on Vehicular Technology},
pages = {1-13},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
3204.
Franchini, Giorgia
GreenNAS: A Green Approach to the Hyperparameters Tuning in Deep Learning Journal Article
In: Mathematics, vol. 12, no. 6, 2024, ISSN: 2227-7390.
@article{math12060850,
title = {GreenNAS: A Green Approach to the Hyperparameters Tuning in Deep Learning},
author = {Giorgia Franchini},
url = {https://www.mdpi.com/2227-7390/12/6/850},
doi = {10.3390/math12060850},
issn = {2227-7390},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {Mathematics},
volume = {12},
number = {6},
abstract = {This paper discusses the challenges of the hyperparameter tuning in deep learning models and proposes a green approach to the neural architecture search process that minimizes its environmental impact. The traditional approach of neural architecture search involves sweeping the entire space of possible architectures, which is computationally expensive and time-consuming. Recently, to address this issue, performance predictors have been proposed to estimate the performance of different architectures, thereby reducing the search space and speeding up the exploration process. The proposed approach aims to develop a performance predictor by training only a small percentage of the possible hyperparameter configurations. The suggested predictor can be queried to find the best configurations without training them on the dataset. Numerical examples of image denoising and classification enable us to evaluate the performance of the proposed approach in terms of performance and time complexity.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
This paper discusses the challenges of the hyperparameter tuning in deep learning models and proposes a green approach to the neural architecture search process that minimizes its environmental impact. The traditional approach of neural architecture search involves sweeping the entire space of possible architectures, which is computationally expensive and time-consuming. Recently, to address this issue, performance predictors have been proposed to estimate the performance of different architectures, thereby reducing the search space and speeding up the exploration process. The proposed approach aims to develop a performance predictor by training only a small percentage of the possible hyperparameter configurations. The suggested predictor can be queried to find the best configurations without training them on the dataset. Numerical examples of image denoising and classification enable us to evaluate the performance of the proposed approach in terms of performance and time complexity.
3203.
Farahani, Rezsa
Random Search as a Baseline for Sparse Neural Network Architecture Search Technical Report
2024.
@techreport{DBLP:journals/corr/abs-2403-08265,
title = {Random Search as a Baseline for Sparse Neural Network Architecture Search},
author = {Rezsa Farahani},
url = {https://doi.org/10.48550/arXiv.2403.08265},
doi = {10.48550/ARXIV.2403.08265},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {CoRR},
volume = {abs/2403.08265},
keywords = {},
pubstate = {published},
tppubtype = {techreport}
}
3202.
Situ, Haozhen; He, Zhimin; Zheng, Shenggen; Li, Lvzhou
Distributed quantum architecture search Technical Report
2024.
@techreport{situ2024distributed,
title = {Distributed quantum architecture search},
author = {Haozhen Situ and Zhimin He and Shenggen Zheng and Lvzhou Li},
url = {https://arxiv.org/abs/2403.06214v1},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
keywords = {},
pubstate = {published},
tppubtype = {techreport}
}
3201.
Yuan, Gonglin
Evolutionary Neural Architecture Search for Image Classification Journal Article
In: 2024.
@article{Yuan2024,
title = {Evolutionary Neural Architecture Search for Image Classification},
author = {Gonglin Yuan},
url = {https://openaccess.wgtn.ac.nz/articles/thesis/Evolutionary_Neural_Architecture_Search_for_Image_Classification/25378900},
doi = {10.26686/wgtn.25378900},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
3200.
Zou, Juan; Chu, Han; Xia, Yizhang; Xu, Junwen; Liu, Yuan; Hou, Zhanglu
Multiple Population Alternate Evolution Neural Architecture Search Technical Report
2024.
@techreport{DBLP:journals/corr/abs-2403-07035,
title = {Multiple Population Alternate Evolution Neural Architecture Search},
author = {Juan Zou and Han Chu and Yizhang Xia and Junwen Xu and Yuan Liu and Zhanglu Hou},
url = {https://doi.org/10.48550/arXiv.2403.07035},
doi = {10.48550/ARXIV.2403.07035},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {CoRR},
volume = {abs/2403.07035},
keywords = {},
pubstate = {published},
tppubtype = {techreport}
}
3199.
Asthana, Rohan; Conrad, Joschua; Dawoud, Youssef; Ortmanns, Maurits; Belagiannis, Vasileios
Multi-conditioned Graph Diffusion for Neural Architecture Search Bachelor Thesis
2024.
@bachelorthesis{DBLP:journals/corr/abs-2403-06020,
title = {Multi-conditioned Graph Diffusion for Neural Architecture Search},
author = {Rohan Asthana and Joschua Conrad and Youssef Dawoud and Maurits Ortmanns and Vasileios Belagiannis},
url = {https://doi.org/10.48550/arXiv.2403.06020},
doi = {10.48550/ARXIV.2403.06020},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {CoRR},
volume = {abs/2403.06020},
keywords = {},
pubstate = {published},
tppubtype = {bachelorthesis}
}
3198.
Xue, Yu; Han, Xiaolong; Neri, Ferrante; Qin, Jiafeng; Pelusi, Danilo
A Gradient-Guided Evolutionary Neural Architecture Search Journal Article
In: IEEE Transactions on Neural Networks and Learning Systems, pp. 1-13, 2024.
@article{10465622,
title = {A Gradient-Guided Evolutionary Neural Architecture Search},
author = {Yu Xue and Xiaolong Han and Ferrante Neri and Jiafeng Qin and Danilo Pelusi},
url = {https://ieeexplore.ieee.org/abstract/document/10465622},
doi = {10.1109/TNNLS.2024.3371432},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {IEEE Transactions on Neural Networks and Learning Systems},
pages = {1-13},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
3197.
Gao, Yuan; ZHANG, WEIZHONG; Luo, Wenhan; Ma, Lin; Yu, Jin-Gang; Xia, Gui-Song; Ma, Jiayi
Free Lunches in Auxiliary Learning: Exploiting Auxiliary Labels with Negligibly Extra Inference Cost Proceedings Article
In: The Twelfth International Conference on Learning Representations, 2024.
@inproceedings{<LineBreak>gao2024free,
title = {Free Lunches in Auxiliary Learning: Exploiting Auxiliary Labels with Negligibly Extra Inference Cost},
author = {Yuan Gao and WEIZHONG ZHANG and Wenhan Luo and Lin Ma and Jin-Gang Yu and Gui-Song Xia and Jiayi Ma},
url = {https://openreview.net/forum?id=cINwAhrgLf},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
booktitle = {The Twelfth International Conference on Learning Representations},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
3196.
Schiessler, Elisabeth J.; Aydin, Roland C.; Cyron, Christian J.
ECToNAS: Evolutionary Cross-Topology Neural Architecture Search Technical Report
2024.
@techreport{DBLP:journals/corr/abs-2403-05123,
title = {ECToNAS: Evolutionary Cross-Topology Neural Architecture Search},
author = {Elisabeth J. Schiessler and Roland C. Aydin and Christian J. Cyron},
url = {https://doi.org/10.48550/arXiv.2403.05123},
doi = {10.48550/ARXIV.2403.05123},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {CoRR},
volume = {abs/2403.05123},
keywords = {},
pubstate = {published},
tppubtype = {techreport}
}
3195.
Xie, Weisheng; Li, Hui; Fang, Xuwei; Li, Shaoyuan
DARTS-PT-CORE: Collaborative and Regularized Perturbation-based Architecture Selection for differentiable NAS Journal Article
In: Neurocomputing, vol. 580, pp. 127522, 2024, ISSN: 0925-2312.
@article{XIE2024127522b,
title = {DARTS-PT-CORE: Collaborative and Regularized Perturbation-based Architecture Selection for differentiable NAS},
author = {Weisheng Xie and Hui Li and Xuwei Fang and Shaoyuan Li},
url = {https://www.sciencedirect.com/science/article/pii/S0925231224002935},
doi = {https://doi.org/10.1016/j.neucom.2024.127522},
issn = {0925-2312},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {Neurocomputing},
volume = {580},
pages = {127522},
abstract = {DARTS-PT is a well-known differentiable NAS method that measures the operation strength through its contribution to the supernet performance, extracting architecture from the underlying supernet. However, persistent issues of degraded architecture in DARTS-PT have been identified in recent studies. In response, we undertake a comprehensive analysis of this performance degradation issue and identify two primary contributing factors: the unfavorable competition among correlated operations during the operation selection process and the unfair advantage of parameter-free operations within DARTS-PT supernet. Building upon these findings, we propose DARTS-PT-CORE, a novel architecture selection algorithm that incorporates a collaborative operation competition mechanism and a regularization technique in the perturbation-based architecture selection approach. Our method aims to mitigate the negative effects of competition among correlated operations, yielding more reliable operation contribution scores. Furthermore, our regularization technique addresses the unfair advantage of parameter-free operations, facilitating a more balanced architecture selection process. Extensive experiments conducted on various datasets and search spaces indicate that DARTS-PT-CORE outperforms other state of-the-art methods. Specifically, in the DARTS search space, DARTS-PT-CORE achieves 2.43% test error on CIFAR10 and 16.23% test error on CIFAR100, while the search time is less than 0.8 GPU days. When transferring to ImageNet, DARTS-PT-CORE achieves 24.97% top-1 error. Such results underscore the effectiveness of our method in enhancing the reliability and balance of architecture selection in differentiable NAS.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
DARTS-PT is a well-known differentiable NAS method that measures the operation strength through its contribution to the supernet performance, extracting architecture from the underlying supernet. However, persistent issues of degraded architecture in DARTS-PT have been identified in recent studies. In response, we undertake a comprehensive analysis of this performance degradation issue and identify two primary contributing factors: the unfavorable competition among correlated operations during the operation selection process and the unfair advantage of parameter-free operations within DARTS-PT supernet. Building upon these findings, we propose DARTS-PT-CORE, a novel architecture selection algorithm that incorporates a collaborative operation competition mechanism and a regularization technique in the perturbation-based architecture selection approach. Our method aims to mitigate the negative effects of competition among correlated operations, yielding more reliable operation contribution scores. Furthermore, our regularization technique addresses the unfair advantage of parameter-free operations, facilitating a more balanced architecture selection process. Extensive experiments conducted on various datasets and search spaces indicate that DARTS-PT-CORE outperforms other state of-the-art methods. Specifically, in the DARTS search space, DARTS-PT-CORE achieves 2.43% test error on CIFAR10 and 16.23% test error on CIFAR100, while the search time is less than 0.8 GPU days. When transferring to ImageNet, DARTS-PT-CORE achieves 24.97% top-1 error. Such results underscore the effectiveness of our method in enhancing the reliability and balance of architecture selection in differentiable NAS.
3194.
Cui, Suhan; Mitra, Prasenjit
Automated Multi-Task Learning for Joint Disease Prediction on Electronic Health Records Technical Report
2024.
@techreport{DBLP:journals/corr/abs-2403-04086,
title = {Automated Multi-Task Learning for Joint Disease Prediction on Electronic Health Records},
author = {Suhan Cui and Prasenjit Mitra},
url = {https://doi.org/10.48550/arXiv.2403.04086},
doi = {10.48550/ARXIV.2403.04086},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {CoRR},
volume = {abs/2403.04086},
keywords = {},
pubstate = {published},
tppubtype = {techreport}
}
3193.
Jin, 2 Jinjie Huang Cong
Enhanced Differentiable Architecture Search Based on Asymptotic Regularization Journal Article
In: Computers, Materials & Continua, vol. 78, no. 2, pp. 1547–1568, 2024, ISSN: 1546-2226.
@article{cmc.2023.047489,
title = {Enhanced Differentiable Architecture Search Based on Asymptotic Regularization},
author = {2 Jinjie Huang Cong Jin},
url = {http://www.techscience.com/cmc/v78n2/55579},
doi = {10.32604/cmc.2023.047489},
issn = {1546-2226},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {Computers, Materials & Continua},
volume = {78},
number = {2},
pages = {1547–1568},
abstract = {In differentiable search architecture search methods, a more efficient search space design can significantly improve the performance of the searched architecture, thus requiring people to carefully define the search space with different complexity according to various operations. Meanwhile rationalizing the search strategies to explore the well-defined search space will further improve the speed and efficiency of architecture search. With this in mind, we propose a faster and more efficient differentiable architecture search method, AllegroNAS. Firstly, we introduce a more efficient search space enriched by the introduction of two redefined convolution modules. Secondly, we utilize a more efficient architectural parameter regularization method, mitigating the overfitting problem during the search process and reducing the error brought about by gradient approximation. Meanwhile, we introduce a natural exponential cosine annealing method to make the learning rate of the neural network training process more suitable for the search procedure. Moreover, group convolution and data augmentation are employed to reduce the computational cost. Finally, through extensive experiments on several public datasets, we demonstrate that our method can more swiftly search for better-performing neural network architectures in a more efficient search space, thus validating the effectiveness of our approach.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
In differentiable search architecture search methods, a more efficient search space design can significantly improve the performance of the searched architecture, thus requiring people to carefully define the search space with different complexity according to various operations. Meanwhile rationalizing the search strategies to explore the well-defined search space will further improve the speed and efficiency of architecture search. With this in mind, we propose a faster and more efficient differentiable architecture search method, AllegroNAS. Firstly, we introduce a more efficient search space enriched by the introduction of two redefined convolution modules. Secondly, we utilize a more efficient architectural parameter regularization method, mitigating the overfitting problem during the search process and reducing the error brought about by gradient approximation. Meanwhile, we introduce a natural exponential cosine annealing method to make the learning rate of the neural network training process more suitable for the search procedure. Moreover, group convolution and data augmentation are employed to reduce the computational cost. Finally, through extensive experiments on several public datasets, we demonstrate that our method can more swiftly search for better-performing neural network architectures in a more efficient search space, thus validating the effectiveness of our approach.
3192.
Chen, Jialin; Cai, Zhiqiang; Xu, Ke; Wu, Di; Cao, Wei
Qubit-Wise Architecture Search Method for Variational Quantum Circuits Technical Report
2024.
@techreport{chen2024qubitwise,
title = {Qubit-Wise Architecture Search Method for Variational Quantum Circuits},
author = {Jialin Chen and Zhiqiang Cai and Ke Xu and Di Wu and Wei Cao},
url = {https://arxiv.org/abs/2403.04268},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
keywords = {},
pubstate = {published},
tppubtype = {techreport}
}
3191.
Wang, Wenna; Ran, Lingyan; Yin, Hanlin; Sun, Mingjun; Zhang, Xiuwei; Zhang, Yanning
Hierarchical Shared Architecture Search for Real-Time Semantic Segmentation of Remote Sensing Images Journal Article
In: IEEE Transactions on Geoscience and Remote Sensing, vol. 62, pp. 1-13, 2024.
@article{10460490,
title = {Hierarchical Shared Architecture Search for Real-Time Semantic Segmentation of Remote Sensing Images},
author = {Wenna Wang and Lingyan Ran and Hanlin Yin and Mingjun Sun and Xiuwei Zhang and Yanning Zhang},
url = {https://ieeexplore.ieee.org/abstract/document/10460490},
doi = {10.1109/TGRS.2024.3373493},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {IEEE Transactions on Geoscience and Remote Sensing},
volume = {62},
pages = {1-13},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
3190.
Ji, Mengfei; Chang, Yuchun; Zhang, Baolin; Al-Ars, Zaid
NASH: Neural Architecture Search for Hardware-Optimized Machine Learning Models Technical Report
2024.
@techreport{DBLP:journals/corr/abs-2403-01845,
title = {NASH: Neural Architecture Search for Hardware-Optimized Machine Learning Models},
author = {Mengfei Ji and Yuchun Chang and Baolin Zhang and Zaid Al-Ars},
url = {https://doi.org/10.48550/arXiv.2403.01845},
doi = {10.48550/ARXIV.2403.01845},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {CoRR},
volume = {abs/2403.01845},
keywords = {},
pubstate = {published},
tppubtype = {techreport}
}
3189.
Zhu, Wenbo; Hu, Yongcong; Zhu, Zhengjun; Yeh, Wei-Chang; Li, Haibing; Zhang, Zhongbo; Fu, Weijie
Searching by Topological Complexity: Lightweight Neural Architecture Search for Coal and Gangue Classification Journal Article
In: Mathematics, vol. 12, no. 5, 2024, ISSN: 2227-7390.
@article{math12050759,
title = {Searching by Topological Complexity: Lightweight Neural Architecture Search for Coal and Gangue Classification},
author = {Wenbo Zhu and Yongcong Hu and Zhengjun Zhu and Wei-Chang Yeh and Haibing Li and Zhongbo Zhang and Weijie Fu},
url = {https://www.mdpi.com/2227-7390/12/5/759},
doi = {10.3390/math12050759},
issn = {2227-7390},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {Mathematics},
volume = {12},
number = {5},
abstract = {Lightweight and adaptive adjustment are key research directions for deep neural networks (DNNs). In coal industry mining, frequent changes in raw coal sources and production batches can cause uneven distribution of appearance features, leading to concept drift problems. The network architecture and parameters should be adjusted frequently to avoid a decline in model accuracy. This poses a significant challenge for those without specialist expertise. Although the Neural Architecture Search (NAS) has a strong ability to automatically generate networks, enabling the automatic design of highly accurate networks, it often comes with complex internal topological connections. These redundant architectures do not always effectively improve network performance, especially in resource-constrained environments, where their computational efficiency is significantly reduced. In this paper, we propose a method called Topology Complexity Neural Architecture Search (TCNAS). TCNAS proposes a new method for evaluating the topological complexity of neural networks and uses both topological complexity and accuracy to guide the search, effectively obtaining lightweight and efficient networks. TCNAS employs an adaptive shrinking search space optimization method, which gradually eliminates poorly performing cells to reduce the search space, thereby improving search efficiency and solving the problem of space explosion. In the classification experiments of coal and gangue, the optimal network designed by TCNAS has an accuracy of 83.3%. And its structure is much simpler, which is about 1/53 of the parameters of the network dedicated to coal and gangue recognition. Experiments have shown that TCNAS is able to generate networks that are both efficient and simple for resource-constrained industrial applications.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Lightweight and adaptive adjustment are key research directions for deep neural networks (DNNs). In coal industry mining, frequent changes in raw coal sources and production batches can cause uneven distribution of appearance features, leading to concept drift problems. The network architecture and parameters should be adjusted frequently to avoid a decline in model accuracy. This poses a significant challenge for those without specialist expertise. Although the Neural Architecture Search (NAS) has a strong ability to automatically generate networks, enabling the automatic design of highly accurate networks, it often comes with complex internal topological connections. These redundant architectures do not always effectively improve network performance, especially in resource-constrained environments, where their computational efficiency is significantly reduced. In this paper, we propose a method called Topology Complexity Neural Architecture Search (TCNAS). TCNAS proposes a new method for evaluating the topological complexity of neural networks and uses both topological complexity and accuracy to guide the search, effectively obtaining lightweight and efficient networks. TCNAS employs an adaptive shrinking search space optimization method, which gradually eliminates poorly performing cells to reduce the search space, thereby improving search efficiency and solving the problem of space explosion. In the classification experiments of coal and gangue, the optimal network designed by TCNAS has an accuracy of 83.3%. And its structure is much simpler, which is about 1/53 of the parameters of the network dedicated to coal and gangue recognition. Experiments have shown that TCNAS is able to generate networks that are both efficient and simple for resource-constrained industrial applications.
3188.
Zou, Juan; Jiang, Weiwei; Xia, Yizhang; Liu, Yuan; Hou, Zhanglu
G-EvoNAS: Evolutionary Neural Architecture Search Based on Network Growth Bachelor Thesis
2024.
@bachelorthesis{DBLP:journals/corr/abs-2403-02667,
title = {G-EvoNAS: Evolutionary Neural Architecture Search Based on Network Growth},
author = {Juan Zou and Weiwei Jiang and Yizhang Xia and Yuan Liu and Zhanglu Hou},
url = {https://doi.org/10.48550/arXiv.2403.02667},
doi = {10.48550/ARXIV.2403.02667},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {CoRR},
volume = {abs/2403.02667},
keywords = {},
pubstate = {published},
tppubtype = {bachelorthesis}
}
3187.
Akhauri, Yash; Abdelfattah, Mohamed S.
On Latency Predictors for Neural Architecture Search Bachelor Thesis
2024.
@bachelorthesis{DBLP:journals/corr/abs-2403-02446,
title = {On Latency Predictors for Neural Architecture Search},
author = {Yash Akhauri and Mohamed S. Abdelfattah},
url = {https://doi.org/10.48550/arXiv.2403.02446},
doi = {10.48550/ARXIV.2403.02446},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {CoRR},
volume = {abs/2403.02446},
keywords = {},
pubstate = {published},
tppubtype = {bachelorthesis}
}
3186.
Xue, Wenyuan; Lu, Yichen; Wang, Zhi; Cao, Shengxian; Sui, Mengxuan; Yang, Yuan; Li, Jiyuan; Xie, Yubin
Reconstructing near-water-wall temperature in coal-fired boilers using improved transfer learning and hidden layer configuration optimization Journal Article
In: Energy, vol. 294, pp. 130860, 2024, ISSN: 0360-5442.
@article{XUE2024130860,
title = {Reconstructing near-water-wall temperature in coal-fired boilers using improved transfer learning and hidden layer configuration optimization},
author = {Wenyuan Xue and Yichen Lu and Zhi Wang and Shengxian Cao and Mengxuan Sui and Yuan Yang and Jiyuan Li and Yubin Xie},
url = {https://www.sciencedirect.com/science/article/pii/S0360544224006327},
doi = {https://doi.org/10.1016/j.energy.2024.130860},
issn = {0360-5442},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {Energy},
volume = {294},
pages = {130860},
abstract = {The temperature field is a critical factor for ensuring the safe combustion and energy conservation in boilers. However, an effective method for reconstructing the temperature field near the water wall is still under exploration. In this paper, a method for online reconstruction of the temperature field distribution near the water wall in a 330 MW tangentially fired coal boiler is proposed, which progresses from a well-established model for the entire furnace. A two-branch fusion network for transfer learning (TBFN-TL) method is proposed, incorporating additional key parameters, heat flux, during the transfer process to enhance the effectiveness. A Bayesian hierarchical neural architecture search (BHNAS) method is proposed to optimize the configuration of the hidden layers in building neural networks. Compared with computational fluid dynamics (CFD) results, the mean absolute percentage error (MAPE) of the reconstruction results for the entire furnace model, traditional transfer learning methods, and the proposed TBFN-TL are 11.57%, 5.738%, and 2.052%, respectively, demonstrating a significant enhancement. The proposed BHNAS method extends the search optimization space, obtaining more excellent configurations for the hidden layer nodes. The proposed methods have significant implications for temperature field reconstruction, the field of transfer learning, and the optimization of hidden layer configurations.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
The temperature field is a critical factor for ensuring the safe combustion and energy conservation in boilers. However, an effective method for reconstructing the temperature field near the water wall is still under exploration. In this paper, a method for online reconstruction of the temperature field distribution near the water wall in a 330 MW tangentially fired coal boiler is proposed, which progresses from a well-established model for the entire furnace. A two-branch fusion network for transfer learning (TBFN-TL) method is proposed, incorporating additional key parameters, heat flux, during the transfer process to enhance the effectiveness. A Bayesian hierarchical neural architecture search (BHNAS) method is proposed to optimize the configuration of the hidden layers in building neural networks. Compared with computational fluid dynamics (CFD) results, the mean absolute percentage error (MAPE) of the reconstruction results for the entire furnace model, traditional transfer learning methods, and the proposed TBFN-TL are 11.57%, 5.738%, and 2.052%, respectively, demonstrating a significant enhancement. The proposed BHNAS method extends the search optimization space, obtaining more excellent configurations for the hidden layer nodes. The proposed methods have significant implications for temperature field reconstruction, the field of transfer learning, and the optimization of hidden layer configurations.
3185.
Xue, Yu; Chen, Kun; Neri, Ferrante
Differentiable Architecture Search with Attention Mechanisms for Generative Adversarial Networks Journal Article
In: IEEE transactions on emerging topics in computational intelligence, 2024, ISSN: 2471-285X.
@article{NeriFerrante2024DASw,
title = {Differentiable Architecture Search with Attention Mechanisms for Generative Adversarial Networks},
author = {Yu Xue and Kun Chen and Ferrante Neri},
url = {https://openresearch.surrey.ac.uk/esploro/outputs/journalArticle/Differentiable-Architecture-Search-with-Attention-Mechanisms/99860066602346?institution=44SUR_INST},
issn = {2471-285X},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {IEEE transactions on emerging topics in computational intelligence},
publisher = {IEEE},
abstract = {—Generative adversarial networks (GANs) are machine learning algorithms that can efficiently generate data such as images. Although GANs are very popular, their training usually lacks stability, with the generator and discriminator networks failing to converge during the training process. To address this problem and improve the stability of GANs, in this paper, we automate the design of stable GANs architectures through a novel approach: differentiable architecture search with attention mechanisms for generative adversarial networks (DAMGAN). We construct a generator supernet and search for the optimal generator network within it. We propose incorporating two attention mechanisms between each pair of nodes in the supernet. The first attention mechanism, down attention, selects the optimal candidate operation of each edge in the supernet, while the second attention mechanism, up attention, improves the training stability of the supernet and limits the computational cost of the search by selecting the most important feature maps for the following candidate operations. Experimental results show that the architectures searched by our method obtain a state-of-the-art inception score (IS) of 8.99 and a very competitive Fréchet inception distance (FID) of 10.27 on the CIFAR-10 dataset. Competitive results were also obtained on the STL-10 dataset (IS = 10.35},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
—Generative adversarial networks (GANs) are machine learning algorithms that can efficiently generate data such as images. Although GANs are very popular, their training usually lacks stability, with the generator and discriminator networks failing to converge during the training process. To address this problem and improve the stability of GANs, in this paper, we automate the design of stable GANs architectures through a novel approach: differentiable architecture search with attention mechanisms for generative adversarial networks (DAMGAN). We construct a generator supernet and search for the optimal generator network within it. We propose incorporating two attention mechanisms between each pair of nodes in the supernet. The first attention mechanism, down attention, selects the optimal candidate operation of each edge in the supernet, while the second attention mechanism, up attention, improves the training stability of the supernet and limits the computational cost of the search by selecting the most important feature maps for the following candidate operations. Experimental results show that the architectures searched by our method obtain a state-of-the-art inception score (IS) of 8.99 and a very competitive Fréchet inception distance (FID) of 10.27 on the CIFAR-10 dataset. Competitive results were also obtained on the STL-10 dataset (IS = 10.35
3184.
Shang, Ronghua; Liu, Hangcheng; Li, Wenzheng; Zhang, Weitong; Ma, Teng; Jiao, Licheng
An efficient evolutionary architecture search for variational autoencoder with alternating optimization and adaptive crossover Journal Article
In: Swarm and Evolutionary Computation, vol. 86, pp. 101520, 2024, ISSN: 2210-6502.
@article{SHANG2024101520,
title = {An efficient evolutionary architecture search for variational autoencoder with alternating optimization and adaptive crossover},
author = {Ronghua Shang and Hangcheng Liu and Wenzheng Li and Weitong Zhang and Teng Ma and Licheng Jiao},
url = {https://www.sciencedirect.com/science/article/pii/S2210650224000531},
doi = {https://doi.org/10.1016/j.swevo.2024.101520},
issn = {2210-6502},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {Swarm and Evolutionary Computation},
volume = {86},
pages = {101520},
abstract = {Variational autoencoder is a commonly unsupervised learning model. However, its complex structure hinders the utilization of the network architecture search algorithm to release researchers from tedious manual design. To design excellent architectures automatically, this paper proposes an efficient evolutionary architecture search for variational autoencoder with alternating optimization and adaptive crossover(AOC-VAE). Firstly, to alleviate the problem of large search space when automatically designing variational autoencoders, AOC-VAE designs an alternating optimized search mechanism based on the specific coupling of encoder and decoder in variational autoencoders, which reduces the original huge search space almost to half. Then, AOC-VAE can find quickly the optimal individual in the solution space by designing an adaptive crossover mechanism. In early evolutionary period, the structural differences between individuals are relatively significant, making crossover operations more inclined to exchange structural information between individuals. As evolution progresses, the individual structures in the population tend to be similar, and the exchange of information concentrates on the parameter. Finally, in the optimization process, a fitness evaluation mechanism based on dynamic weights is designed to accurately find out the outstanding individuals under the current optimization goal. Individual fitness in the population is more inclined to be affected by the current optimization goal, thus guiding the population to evolve according to the optimization goal at different stages. AOC-VAE is verified on MNIST, SVHN, CIFAR-10, and CIFAR-100 benchmark datasets and compared with 14 algorithms. The experimental results show that the VAE network structure designed by the AOC-VAE performs well in the image classification task.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Variational autoencoder is a commonly unsupervised learning model. However, its complex structure hinders the utilization of the network architecture search algorithm to release researchers from tedious manual design. To design excellent architectures automatically, this paper proposes an efficient evolutionary architecture search for variational autoencoder with alternating optimization and adaptive crossover(AOC-VAE). Firstly, to alleviate the problem of large search space when automatically designing variational autoencoders, AOC-VAE designs an alternating optimized search mechanism based on the specific coupling of encoder and decoder in variational autoencoders, which reduces the original huge search space almost to half. Then, AOC-VAE can find quickly the optimal individual in the solution space by designing an adaptive crossover mechanism. In early evolutionary period, the structural differences between individuals are relatively significant, making crossover operations more inclined to exchange structural information between individuals. As evolution progresses, the individual structures in the population tend to be similar, and the exchange of information concentrates on the parameter. Finally, in the optimization process, a fitness evaluation mechanism based on dynamic weights is designed to accurately find out the outstanding individuals under the current optimization goal. Individual fitness in the population is more inclined to be affected by the current optimization goal, thus guiding the population to evolve according to the optimization goal at different stages. AOC-VAE is verified on MNIST, SVHN, CIFAR-10, and CIFAR-100 benchmark datasets and compared with 14 algorithms. The experimental results show that the VAE network structure designed by the AOC-VAE performs well in the image classification task.
3183.
Rahman, Md Hafizur; Chakraborty, Prabuddha
LeMo-NADe: Multi-Parameter Neural Architecture Discovery with LLMs Technical Report
2024.
@techreport{DBLP:journals/corr/abs-2402-18443b,
title = {LeMo-NADe: Multi-Parameter Neural Architecture Discovery with LLMs},
author = {Md Hafizur Rahman and Prabuddha Chakraborty},
url = {https://doi.org/10.48550/arXiv.2402.18443},
doi = {10.48550/ARXIV.2402.18443},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {CoRR},
volume = {abs/2402.18443},
keywords = {},
pubstate = {published},
tppubtype = {techreport}
}
3182.
Garavagno, Andrea Mattia; Ragusa, Edoardo; Frisoli, Antonio; Gastaldo, Paolo
Running hardware-aware neural architecture search on embedded devices under 512MB of RAM Proceedings Article
In: 2024 IEEE International Conference on Consumer Electronics (ICCE), pp. 1-2, 2024.
@inproceedings{10444268,
title = {Running hardware-aware neural architecture search on embedded devices under 512MB of RAM},
author = {Andrea Mattia Garavagno and Edoardo Ragusa and Antonio Frisoli and Paolo Gastaldo},
url = {https://ieeexplore.ieee.org/abstract/document/10444268},
doi = {10.1109/ICCE59016.2024.10444268},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
booktitle = {2024 IEEE International Conference on Consumer Electronics (ICCE)},
pages = {1-2},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
3181.
Gambella, Matteo; Pittorino, Fabrizio; Roveri, Manuel
FlatNAS: optimizing Flatness in Neural Architecture Search for Out-of-Distribution Robustness Technical Report
2024.
@techreport{DBLP:journals/corr/abs-2402-19102,
title = {FlatNAS: optimizing Flatness in Neural Architecture Search for Out-of-Distribution Robustness},
author = {Matteo Gambella and Fabrizio Pittorino and Manuel Roveri},
url = {https://doi.org/10.48550/arXiv.2402.19102},
doi = {10.48550/ARXIV.2402.19102},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {CoRR},
volume = {abs/2402.19102},
keywords = {},
pubstate = {published},
tppubtype = {techreport}
}
3180.
Wu, Xiang; Zhang, Yong-Ting; Lai, Khin-Wee; Yang, Ming-Zhao; Yang, Ge-Lan; Wang, Huan-Huan
A Novel Centralized Federated Deep Fuzzy Neural Network with Multi-objectives Neural Architecture Search for Epistatic Detection Journal Article
In: IEEE Transactions on Fuzzy Systems, pp. 1-13, 2024.
@article{10445010,
title = {A Novel Centralized Federated Deep Fuzzy Neural Network with Multi-objectives Neural Architecture Search for Epistatic Detection},
author = {Xiang Wu and Yong-Ting Zhang and Khin-Wee Lai and Ming-Zhao Yang and Ge-Lan Yang and Huan-Huan Wang},
url = {https://ieeexplore.ieee.org/abstract/document/10445010},
doi = {10.1109/TFUZZ.2024.3369944},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {IEEE Transactions on Fuzzy Systems},
pages = {1-13},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
3179.
Gomez-Rosero, Santiago; Capretz, Miriam A. M.
Anomaly detection in time-series data using evolutionary neural architecture search with non-differentiable functions Journal Article
In: Applied Soft Computing, vol. 155, pp. 111442, 2024, ISSN: 1568-4946.
@article{GOMEZROSERO2024111442,
title = {Anomaly detection in time-series data using evolutionary neural architecture search with non-differentiable functions},
author = {Santiago Gomez-Rosero and Miriam A. M. Capretz},
url = {https://www.sciencedirect.com/science/article/pii/S1568494624002163},
doi = {https://doi.org/10.1016/j.asoc.2024.111442},
issn = {1568-4946},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {Applied Soft Computing},
volume = {155},
pages = {111442},
abstract = {Deep neural networks have become the benchmark in diverse fields such as energy consumption forecasting, speech recognition, and anomaly detection, owing to their ability to efficiently process and analyse data. However, they face challenges in managing the complexity and variability in time series data, often leading to increased model complexity and prolonged search duration during parameter tuning. This paper proposes a novel anomaly detection approach through evolutionary neural architecture search (AD-ENAS), which is specifically designed for time series data. The proposed approach focuses on the search for the optimal and minimal neural network architecture. The AD-ENAS method consists of two main phases: architecture evolution and weight adjustment. The architecture evolution phase highlights the importance of neural network architecture by evaluating the fitness of each network agent using shared weight values. Subsequently, the convolutional matrix adaptation technique is used in the next phase for optimal weight adjustment of the neural network. The proposed AD-ENAS method operates without relying on differentiable functions, thus expanding the scope of neural network design beyond traditional backpropagation-based approaches. Various non-differentiable loss functions are explored to facilitate effective architecture search and weight adjustment. Comparative experiments are conducted with five baseline anomaly detection methods on three well-known datasets from reputable sources such as NASA SMAP, NASA MSL and Yahoo S5-A1. The results demonstrate that the AD-ENAS approach effectively evolves neural network architectures, outperforming baseline methods with F1 scores across the three datasets (MSL: 0.942, SMAP: 0.961, Yahoo S5-A1: 0.988) with non-differentiable loss functions, showcasing its efficacy in detecting anomalies in time series data.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Deep neural networks have become the benchmark in diverse fields such as energy consumption forecasting, speech recognition, and anomaly detection, owing to their ability to efficiently process and analyse data. However, they face challenges in managing the complexity and variability in time series data, often leading to increased model complexity and prolonged search duration during parameter tuning. This paper proposes a novel anomaly detection approach through evolutionary neural architecture search (AD-ENAS), which is specifically designed for time series data. The proposed approach focuses on the search for the optimal and minimal neural network architecture. The AD-ENAS method consists of two main phases: architecture evolution and weight adjustment. The architecture evolution phase highlights the importance of neural network architecture by evaluating the fitness of each network agent using shared weight values. Subsequently, the convolutional matrix adaptation technique is used in the next phase for optimal weight adjustment of the neural network. The proposed AD-ENAS method operates without relying on differentiable functions, thus expanding the scope of neural network design beyond traditional backpropagation-based approaches. Various non-differentiable loss functions are explored to facilitate effective architecture search and weight adjustment. Comparative experiments are conducted with five baseline anomaly detection methods on three well-known datasets from reputable sources such as NASA SMAP, NASA MSL and Yahoo S5-A1. The results demonstrate that the AD-ENAS approach effectively evolves neural network architectures, outperforming baseline methods with F1 scores across the three datasets (MSL: 0.942, SMAP: 0.961, Yahoo S5-A1: 0.988) with non-differentiable loss functions, showcasing its efficacy in detecting anomalies in time series data.
3178.
Rahman, Md Hafizur; Chakraborty, Prabuddha
LeMo-NADe: Multi-Parameter Neural Architecture Discovery with LLMs Technical Report
2024.
@techreport{DBLP:journals/corr/abs-2402-18443,
title = {LeMo-NADe: Multi-Parameter Neural Architecture Discovery with LLMs},
author = {Md Hafizur Rahman and Prabuddha Chakraborty},
url = {https://doi.org/10.48550/arXiv.2402.18443},
doi = {10.48550/ARXIV.2402.18443},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {CoRR},
volume = {abs/2402.18443},
keywords = {},
pubstate = {published},
tppubtype = {techreport}
}
3177.
Zhang, Pengyu; Zhou, Yingbo; Hu, Ming; Feng, Junxian; Weng, Jiawen; Chen, Mingsong
Personalized Federated Instruction Tuning via Neural Architecture Search Technical Report
2024.
@techreport{DBLP:journals/corr/abs-2402-16919,
title = {Personalized Federated Instruction Tuning via Neural Architecture Search},
author = {Pengyu Zhang and Yingbo Zhou and Ming Hu and Junxian Feng and Jiawen Weng and Mingsong Chen},
url = {https://doi.org/10.48550/arXiv.2402.16919},
doi = {10.48550/ARXIV.2402.16919},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {CoRR},
volume = {abs/2402.16919},
keywords = {},
pubstate = {published},
tppubtype = {techreport}
}
3176.
Sukthanker, Rhea Sanjay; Zela, Arber; Staffler, Benedikt; Dooley, Samuel; Grabocka, Josif; Hutter, Frank
Multi-objective Differentiable Neural Architecture Search Technical Report
2024.
@techreport{DBLP:journals/corr/abs-2402-18213,
title = {Multi-objective Differentiable Neural Architecture Search},
author = {Rhea Sanjay Sukthanker and Arber Zela and Benedikt Staffler and Samuel Dooley and Josif Grabocka and Frank Hutter},
url = {https://doi.org/10.48550/arXiv.2402.18213},
doi = {10.48550/ARXIV.2402.18213},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {CoRR},
volume = {abs/2402.18213},
keywords = {},
pubstate = {published},
tppubtype = {techreport}
}
3175.
Gao, Peng; Liu, Xiao; Wang, Yu; Yuan, Ru-Yue
Searching a Lightweight Network Architecture for Thermal Infrared Pedestrian Tracking Technical Report
2024.
@techreport{DBLP:journals/corr/abs-2402-16570,
title = {Searching a Lightweight Network Architecture for Thermal Infrared Pedestrian Tracking},
author = {Peng Gao and Xiao Liu and Yu Wang and Ru-Yue Yuan},
url = {https://doi.org/10.48550/arXiv.2402.16570},
doi = {10.48550/ARXIV.2402.16570},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {CoRR},
volume = {abs/2402.16570},
keywords = {},
pubstate = {published},
tppubtype = {techreport}
}
3174.
Zhang, Rui; Zhang, Peng-Yun; Gao, Mei-Rong; Ma, Jian-Zhe; Pan, Li-Hu
Low-cost architecture performance evaluation strategy based on pixel difference degree contrast measurement Journal Article
In: Applied Soft Computing, vol. 155, pp. 111440, 2024, ISSN: 1568-4946.
@article{ZHANG2024111440,
title = {Low-cost architecture performance evaluation strategy based on pixel difference degree contrast measurement},
author = {Rui Zhang and Peng-Yun Zhang and Mei-Rong Gao and Jian-Zhe Ma and Li-Hu Pan},
url = {https://www.sciencedirect.com/science/article/pii/S156849462400214X},
doi = {https://doi.org/10.1016/j.asoc.2024.111440},
issn = {1568-4946},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {Applied Soft Computing},
volume = {155},
pages = {111440},
abstract = {The time and effort required to manually design deep neural architectures is extremely high, which has led to the development of neural architecture search technology as an automatic architecture design method. However, the neural architecture search convergence process is slow and expensive, and the process requires training a large number of candidate architectures to get the final result. If the final accuracy of an architecture can be predicted from its initial state, this problem can be greatly alleviated. Therefore, this paper proposes a low-cost architecture performance evaluation strategy based on pixel difference degree contrast measurement, which takes 1) the difference matrix value between the feature map generated in the untrained architecture and the original image, and 2) the predicted accuracy of the neural network as evaluation indices. A new multi-index weight comprehensive measurement strategy was introduced to comprehensively score the multi-index, the real architecture performance can be approximately represented by score, which greatly reduces the cost of architecture evaluation. The experimental show that the proposed scoring strategy is highly correlated with real architecture accuracy. In the practical engineering application research, this strategy can search a high-performance architecture with an accuracy of 96.2% within 343.3 s, which proves that the proposed strategy can significantly improve the search efficiency in practical applications, reduce the subjectivity of artificial architecture design, and promote the application of practical time-consuming projects.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
The time and effort required to manually design deep neural architectures is extremely high, which has led to the development of neural architecture search technology as an automatic architecture design method. However, the neural architecture search convergence process is slow and expensive, and the process requires training a large number of candidate architectures to get the final result. If the final accuracy of an architecture can be predicted from its initial state, this problem can be greatly alleviated. Therefore, this paper proposes a low-cost architecture performance evaluation strategy based on pixel difference degree contrast measurement, which takes 1) the difference matrix value between the feature map generated in the untrained architecture and the original image, and 2) the predicted accuracy of the neural network as evaluation indices. A new multi-index weight comprehensive measurement strategy was introduced to comprehensively score the multi-index, the real architecture performance can be approximately represented by score, which greatly reduces the cost of architecture evaluation. The experimental show that the proposed scoring strategy is highly correlated with real architecture accuracy. In the practical engineering application research, this strategy can search a high-performance architecture with an accuracy of 96.2% within 343.3 s, which proves that the proposed strategy can significantly improve the search efficiency in practical applications, reduce the subjectivity of artificial architecture design, and promote the application of practical time-consuming projects.
3173.
Risso, Matteo; Daghero, Francesco; Motetti, Beatrice Alessandra; Pagliari, Daniele Jahier; Macii, Enrico; Poncino, Massimo; Burrello, Alessio
Optimized Deployment of Deep Neural Networks for Visual Pose Estimation on Nano-drones Technical Report
2024.
@techreport{DBLP:journals/corr/abs-2402-15273,
title = {Optimized Deployment of Deep Neural Networks for Visual Pose Estimation on Nano-drones},
author = {Matteo Risso and Francesco Daghero and Beatrice Alessandra Motetti and Daniele Jahier Pagliari and Enrico Macii and Massimo Poncino and Alessio Burrello},
url = {https://doi.org/10.48550/arXiv.2402.15273},
doi = {10.48550/ARXIV.2402.15273},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {CoRR},
volume = {abs/2402.15273},
keywords = {},
pubstate = {published},
tppubtype = {techreport}
}
3172.
Garg, Manav S.
Automated Machine Learning: Evaluation without Training Journal Article
In: 2024.
@article{Garg_2024,
title = {Automated Machine Learning: Evaluation without Training},
author = {Manav S. Garg},
url = {http://dx.doi.org/10.36227/techrxiv.170595826.64565617/v1},
doi = {10.36227/techrxiv.170595826.64565617/v1},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
publisher = {Institute of Electrical and Electronics Engineers (IEEE)},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
3171.
Huang, Hongtao; Chang, Xiaojun; Hu, Wen; Yao, Lina
MatchNAS: Optimizing Edge AI in Sparse-Label Data Contexts via Automating Deep Neural Network Porting for Mobile Deployment Technical Report
2024.
@techreport{DBLP:journals/corr/abs-2402-13525,
title = {MatchNAS: Optimizing Edge AI in Sparse-Label Data Contexts via Automating Deep Neural Network Porting for Mobile Deployment},
author = {Hongtao Huang and Xiaojun Chang and Wen Hu and Lina Yao},
url = {https://doi.org/10.48550/arXiv.2402.13525},
doi = {10.48550/ARXIV.2402.13525},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {CoRR},
volume = {abs/2402.13525},
keywords = {},
pubstate = {published},
tppubtype = {techreport}
}
3170.
Cao, Tue Minh; Tran, Nhat Hong; Pham, Hieu H.; Nguyen, Hung Thanh; Nguyen, Le P.
MSTAR: Multi-Scale Backbone Architecture Search for Timeseries Classification Technical Report
2024.
@techreport{DBLP:journals/corr/abs-2402-13822,
title = {MSTAR: Multi-Scale Backbone Architecture Search for Timeseries Classification},
author = {Tue Minh Cao and Nhat Hong Tran and Hieu H. Pham and Hung Thanh Nguyen and Le P. Nguyen},
url = {https://doi.org/10.48550/arXiv.2402.13822},
doi = {10.48550/ARXIV.2402.13822},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {CoRR},
volume = {abs/2402.13822},
keywords = {},
pubstate = {published},
tppubtype = {techreport}
}
3169.
Putra, Rachmad Vidya Wicaksana; Shafique, Muhammad
SpikeNAS: A Fast Memory-Aware Neural Architecture Search Framework for Spiking Neural Network Systems Technical Report
2024.
@techreport{DBLP:journals/corr/abs-2402-11322b,
title = {SpikeNAS: A Fast Memory-Aware Neural Architecture Search Framework for Spiking Neural Network Systems},
author = {Rachmad Vidya Wicaksana Putra and Muhammad Shafique},
url = {https://doi.org/10.48550/arXiv.2402.11322},
doi = {10.48550/ARXIV.2402.11322},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {CoRR},
volume = {abs/2402.11322},
keywords = {},
pubstate = {published},
tppubtype = {techreport}
}
3168.
Bouzidi, Halima; Niar, Sma"ıl; Ouarnoughi, Hamza; Talbi, El-Ghazali
SONATA: Self-adaptive Evolutionary Framework for Hardware-aware Neural Architecture Search Technical Report
2024.
@techreport{DBLP:journals/corr/abs-2402-13204,
title = {SONATA: Self-adaptive Evolutionary Framework for Hardware-aware Neural Architecture Search},
author = {Halima Bouzidi and Sma"ıl Niar and Hamza Ouarnoughi and El-Ghazali Talbi},
url = {https://doi.org/10.48550/arXiv.2402.13204},
doi = {10.48550/ARXIV.2402.13204},
year = {2024},
date = {2024-01-01},
urldate = {2024-01-01},
journal = {CoRR},
volume = {abs/2402.13204},
keywords = {},
pubstate = {published},
tppubtype = {techreport}
}