Literature on Neural Architecture Search

@article{Franchini2023,

title = {Neural architecture search via standard machine learning methodologies},

author = {Giorgia Franchini and Valeria Ruggiero and Federica Porta and Luca Zanni},

url = {https://www.aimspress.com/article/doi/10.3934/mine.2023012?viewType=HTML},

year  = {2023},

date = {2023-01-01},

urldate = {2023-01-01},

journal = {Mathematics in Engineering},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{Wang2022,

title = {Architecture evolution of convolutional neural network using monarch butterfly optimization},

author = {Yong Wang and Xiaobin Qiao and Gai-Ge Wang },

url = {https://link.springer.com/article/10.1007/s12652-022-03766-4},

year  = {2022},

date = {2022-03-01},

urldate = {2022-03-01},

journal = {Journal of Ambient Intelligence and Humanized Computing },

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{10.1145/3486618,

title = {Enable Deep Learning on Mobile Devices: Methods, Systems, and Applications},

author = {Han Cai and Ji Lin and Yujun Lin and Zhijian Liu and Haotian Tang and Hanrui Wang and Ligeng Zhu and Song Han},

url = {https://doi.org/10.1145/3486618},

doi = {10.1145/3486618},

issn = {1084-4309},

year  = {2022},

date = {2022-03-01},

urldate = {2022-03-01},

journal = {ACM Trans. Des. Autom. Electron. Syst.},

volume = {27},

number = {3},

publisher = {Association for Computing Machinery},

address = {New York, NY, USA},

abstract = {Deep neural networks (DNNs) have achieved unprecedented success in the field of artificial intelligence (AI), including computer vision, natural language processing, and speech recognition. However, their superior performance comes at the considerable cost of computational complexity, which greatly hinders their applications in many resource-constrained devices, such as mobile phones and Internet of Things (IoT) devices. Therefore, methods and techniques that are able to lift the efficiency bottleneck while preserving the high accuracy of DNNs are in great demand to enable numerous edge AI applications. This article provides an overview of efficient deep learning methods, systems, and applications. We start from introducing popular model compression methods, including pruning, factorization, quantization, as well as compact model design. To reduce the large design cost of these manual solutions, we discuss the AutoML framework for each of them, such as neural architecture search (NAS) and automated pruning and quantization. We then cover efficient on-device training to enable user customization based on the local data on mobile devices. Apart from general acceleration techniques, we also showcase several task-specific accelerations for point cloud, video, and natural language processing by exploiting their spatial sparsity and temporal/token redundancy. Finally, to support all these algorithmic advancements, we introduce the efficient deep learning system design from both software and hardware perspectives.},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{Jia2022,

title = {Compression of Deep Learning Models for Resource-Constrained Devices},

author = {Liang Jia and Ye Tian and Junguo Zhang},

url = {https://www.hindawi.com/journals/cin/2022/8615374/},

year  = {2022},

date = {2022-02-02},

urldate = {2022-02-02},

journal = {Computational Intelligence and Neuroscience},

volume = {2022},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{GuoEC2022,

title = {Stochastic deep collocation method based on neural architecture search and transfer learning for heterogeneous porous media},

author = {Hongwei Guo and Xiaoying Zhuang and Pengwan Chen and Naif Alajlan and Timon Rabczuk 



},

url = {https://doi.org/10.1007/s00366-021-01586-2},

year  = {2022},

date = {2022-02-01},

urldate = {2022-02-01},

journal = {Engineering with Computers },

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{Yao2022,

title = {Relational structure predictive neural architecture search for multimodal fusion},

author = {Xiao Yao and Fang Li and Yifeng Zeng 

},

url = {https://link.springer.com/article/10.1007/s00500-022-06772-y},

year  = {2022},

date = {2022-02-01},

urldate = {2022-02-01},

journal = {Soft Computing},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{10.1145/3517130,

title = {Efficient Layout Hotspot Detection via Neural Architecture Search},

author = {Yiyang Jiang and Fan Yang and Bei Yu and Dian Zhou and Xuan Zeng},

url = {https://doi.org/10.1145/3517130},

doi = {10.1145/3517130},

issn = {1084-4309},

year  = {2022},

date = {2022-02-01},

urldate = {2022-02-01},

journal = {ACM Trans. Des. Autom. Electron. Syst.},

publisher = {Association for Computing Machinery},

address = {New York, NY, USA},

abstract = {Layout hotspot detection is of great importance in the physical verification flow. Deep neural network models have been applied to hotspot detection and achieved great success. Despite their success, high-performance neural networks are still quite difficult to design. In this paper, we propose a bayesian optimization-based neural architecture search scheme to automatically do this time-consuming and fiddly job. Experimental results on ICCAD 2012 and ICCAD 2019 Contest benchmarks show that the architectures designed by our proposed scheme achieve higher performance on hotspot detection task compared with state-of-the-art manually designed neural networks.},

note = {Just Accepted},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{Hu2022,

title = {Image-based neural architecture automatic search method for hyperspectral image classification},

author = {Zhonggang Hu and Wenxing Bao and Kewen Qu and Hongbo Liang

},

url = {https://www.spiedigitallibrary.org/journals/journal-of-applied-remote-sensing/volume-16/issue-01/016501/Image-based-neural-architecture-automatic-search-method-for-hyperspectral-image/10.1117/1.JRS.16.016501.full?tab=ArticleLinkFigureTable},

year  = {2022},

date = {2022-01-06},

journal = {J. of Applied Remote Sensing},

volume = {16},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{DiaoPANAS2022,

title = {PA-NAS: Partial operation activation for memory-efficient architecture search},

author = {Huabin Diao and Gongyan Li and Shaoyun Xu and Yuexing Hao},

url = {https://link.springer.com/article/10.1007/s10489-021-02961-8},

year  = {2022},

date = {2022-01-05},

urldate = {2022-01-05},

journal = {Applied Intelligence},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{Klos2022,

title = {Neural Architecture Search based on Genetic Algorithm and Deployed in a Bare-Metal Kubernetes Cluster},

author = {Andreas Klos and Marius Rosenbaum and Wolfram Schiffmann},

url = {http://www.ijnc.org/index.php/ijnc/article/view/276},

year  = {2022},

date = {2022-01-03},

urldate = {2022-01-03},

journal = {International Journal of Networking and Computing},

volume = {12},

number = {1},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{FEI2022107410,

title = {Non-technical losses detection using missing values’ pattern and neural architecture search},

author = {Ke Fei and Qi Li and Congcong Zhu},

url = {https://www.sciencedirect.com/science/article/pii/S0142061521006499},

doi = {https://doi.org/10.1016/j.ijepes.2021.107410},

issn = {0142-0615},

year  = {2022},

date = {2022-01-01},

journal = {International Journal of Electrical Power & Energy Systems},

volume = {134},

pages = {107410},

abstract = {The fast growth of non-technical loss (NTL) has gradually become one of the main concerns for distribution network operators (DNOs). Electricity theft which constitutes the main part of NTL not only brings losses to the DNOs, but also reduces the quality of the supply. A traditional detection method relies on utility workers’ experience and consumes a large amount of manpower. Thanks to the emerging of advanced metering infrastructure (AMI), utility companies can now collect detailed data reflecting consumers’ electricity usage, which enabled algorithms-based non-technical loss detection. The current data-based methods focus on the characteristics of electricity consumption thereby less efficient when dealing with rapidly changed electricity theft techniques. This article introduced a new data set, the location information of missing values, to improve the accuracy of non-technical loss detection. The relationship between missing values and electricity theft techniques is analyzed and a neural network model is built through neural architecture search (NAS). The improved model achieved an excellent Area Under Curve (AUC) value around 0.926 which verified the close link between missing values and electricity theft techniques. The nature of neural architecture search allows automatic model update which makes it a user-friendly tool even for engineers without any neural network expertise. A case study was carried out in which the missing value pattern was analyzed through Density-Based Spatial Clustering of Applications with Noise (DBSCAN) clustering algorithm.},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{NISTOR2022115945,

title = {IntelliSwAS: Optimizing deep neural network architectures using a particle swarm-based approach},

author = {Sergiu Cosmin Nistor and Gabriela Czibula},

url = {https://www.sciencedirect.com/science/article/pii/S0957417421012987},

doi = {https://doi.org/10.1016/j.eswa.2021.115945},

issn = {0957-4174},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {Expert Systems with Applications},

volume = {187},

pages = {115945},

abstract = {Deep learning is in a continuous evolution and many domains benefit from this substantial progress in the development of intelligent solutions. While this progress has been swift, there are more and more applications and the specific requirements of each application domain entails much work done by researchers to adapt existing models or create new ones. This traditional approach has produced many successful designs, but recently, automated methods of finding neural network architectures emerged. We introduce IntelliSwAS approach for optimizing deep neural network architectures for a classification or regression task. Image classification was selected as the task, but IntelliSwAS is generic enough such that it could be successfully applied on other classification or regression problems. A particle swarm-based optimization algorithm is proposed for the automatic search for convolutional neural network architectures. The search technique is enhanced by a machine learning model (DAGRNN) which we designed for predicting the quality of the network architectures and thus increasing the performance of the algorithm. The proposed model is able to process data structured as directed acyclic graphs in general and we applied it specifically on network architectures. The network architecture that we discovered using IntelliSwAS surpassed 89.8% of the competing image classification models that we considered for comparison.},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{CHEN2022109389,

title = {Prediction of dMRI signals with neural architecture search},

author = {Haoze Chen and Zhijie Zhang and Mingwu Jin and Fengxiang Wang},

url = {https://www.sciencedirect.com/science/article/pii/S0165027021003241},

doi = {https://doi.org/10.1016/j.jneumeth.2021.109389},

issn = {0165-0270},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {Journal of Neuroscience Methods},

volume = {365},

pages = {109389},

abstract = {Background 

There is growing interest in the neuroscience community in estimating and mapping microscopic properties of brain tissue non-invasively using magnetic resonance measurements. Machine learning methods are actively investigated to predict the signals measured in diffusion magnetic resonance imaging (dMRI). 

New method 

We applied the neural architecture search (NAS) to train a recurrent neural network to generate a multilayer perceptron to predict the dMRI data of unknown signals based on the different acquisition parameters and training data. The search space of NAS is the number of neurons in each layer of the multilayer perceptron network. To our best knowledge, this is the first time to apply NAS to solve the dMRI signal prediction problem. 

Results 

The experimental results demonstrate that the proposed NAS method can achieve fast training and predict dMRI signals accurately. For dMRI signals with four acquisition strategies of double diffusion encoding (DDE), double oscillating diffusion encoding (DODE), multi-shell and DSI-like pulsed gradient spin-echo (PGSE), the mean squared errors of the multilayer perceptron network designed by NAS are 0.0043, 0.0034, 0.0147 and 0.0199, respectively. 

Comparison with existing method(s) 

We also compared NAS with other machine learning prediction methods, such as support vector regression (SVR), decision tree (DT) and random forest (RF), k-nearest neighbors (KNN), adaboost regressor (AR), gradient boosting regressor (GBR) and extra-trees regressor (ET). NAS achieved the better prediction performance in most cases. 

Conclusion 

In this study, NAS was developed for the prediction of dMRI signals and could become an effective prediction tool.},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{CHENG2022107648,

title = {Exploring more diverse network architectures for single image super-resolution},

author = {Guoan Cheng and Ai Matsune and Hao Du and XinZhi Liu and Shu Zhan},

url = {https://www.sciencedirect.com/science/article/pii/S0950705121009102},

doi = {https://doi.org/10.1016/j.knosys.2021.107648},

issn = {0950-7051},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {Knowledge-Based Systems},

volume = {235},

pages = {107648},

abstract = {We propose a plug-and-play neural architecture search (NAS) method to explore diverse architectures for single image super-resolution (SISR). Unlike current NAS-based methods with the single path setting and pipeline setting, our proposed method achieves the trade-off between diverse network architectures and search cost. Our proposed method formulates the task in a differentiable manner, which inherits the architecture parameter optimization method from Discrete Stochastic Neural Architecture Search (DSNAS). Besides the straightforward searching of operations, we also search each node in a cell for the activation function, from-node, and skip-connection node, which diverse the searched architecture topologies. The individually searching of skip-connection node avoids skip-connection excessive phenomenon. Moreover, to alleviate the influence of inconsistent architecture between training and testing periods, we introduce random variables into the architecture parameter as regularization. Benchmark experiments show our state-of-the-art performance under specific parameters and FLOPs constraints. Compared with other NAS-based SISR methods, our proposed methods achieve better performance with less searching time and resources. The superior results further demonstrate the effectiveness of our proposed NAS methods.},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{LI2022116027,

title = {One-shot neural architecture search for fault diagnosis using vibration signals},

author = {Xudong Li and Jianhua Zheng and Mingtao Li and Wenzhen Ma and Yang Hu},

url = {https://www.sciencedirect.com/science/article/pii/S0957417421013737},

doi = {https://doi.org/10.1016/j.eswa.2021.116027},

issn = {0957-4174},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {Expert Systems with Applications},

volume = {190},

pages = {116027},

abstract = {Machine learning method has been widely applied in industrial fault diagnosis, especially the deep learning method. In the field of industrial fault diagnosis, deep learning is mostly used to extract features of vibration signals to achieve end-to-end fault diagnosis systems. Due to the complexity and variety of actual industrial datasets, some deep learning models are designed to be complicated. However, designing neural network architectures requires rich professional knowledge, experience, and a large number of experiments, increasing the difficulty of developing deep learning models. Fortunately, Neural Architecture Search (NAS), a branch of Automated Machine Learning (AutoML), is developing rapidly. Given a search space, NAS can search for networks that perform better than manually designed. In this paper, a one-shot NAS method for fault diagnosis is proposed. The one-shot model is a supernet that contains all candidate networks in a given search space. The supernet is trained to evaluate the actual performance of candidate networks by measuring the difference between its output probability and the true labels. According to the prediction of supernet, the networks with excellent performance can be searched, using some common search methods such as random search or evolutionary algorithm. Finally, the searched network is trained by reusing the weights of the supernet. To evaluate the proposed method, two search spaces are designed, ResNet and Inception search spaces, to search on PHM 2009 Data Challenge gearbox dataset. The state-of-the-art results are obtained, and accuracies of searched ResNet-A and Inception-A are 84.11% and 83.81%, which are 3.29% and 10.88% higher than Reinforcement Learning based NAS.},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{XUE2022108474,

title = {Automated search space and search strategy selection for AutoML},

author = {Chao Xue and Mengting Hu and Xueqi Huang and Chun-Guang Li},

url = {https://www.sciencedirect.com/science/article/pii/S0031320321006506},

doi = {https://doi.org/10.1016/j.patcog.2021.108474},

issn = {0031-3203},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {Pattern Recognition},

volume = {124},

pages = {108474},

abstract = {Existing works on Automated Machine Learning (AutoML) are mainly based on predefined search space. This paper seeks synergetic automation of two ingredients, i.e., search space and search strategies. Specifically, we formulate the automation of search space and search strategies as a combinatorial optimization problem. Our empirical study on many architecture benchmarks shows that identifying the suitable search space exerts more effect than choosing a sophisticated search strategy. Motivated by this, we attempt to leverage a machine learning method to solve the discrete optimization problem, and thus develop a Layered Architecture Search Tree (LArST) approach to synergize these two components. In addition, we use a probe model-based method to extract dataset-wise features, i.e., meta-features, which is able to facilitate the estimation of proper search space and search strategy for a given task. Experimental results show the efficacy of our approach under different search mechanisms and various datasets and hardware platforms.},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{9645440,

title = {Graph neural architecture search: A survey},

author = {Babatounde Moctard Oloulade and Jianliang Gao and Jiamin Chen and Tengfei Lyu and Raeed Al-Sabri},

url = {https://ieeexplore.ieee.org/document/9645440},

doi = {10.26599/TST.2021.9010057},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {Tsinghua Science and Technology},

volume = {27},

number = {4},

pages = {692-708},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{BAYMURZINA202282,

title = {A review of neural architecture search},

author = {Dilyara Baymurzina and Eugene Golikov and Mikhail Burtsev},

url = {https://www.sciencedirect.com/science/article/pii/S0925231221018439},

doi = {https://doi.org/10.1016/j.neucom.2021.12.014},

issn = {0925-2312},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {Neurocomputing},

volume = {474},

pages = {82-93},

abstract = {Despite the impressive progress in neural network architecture design, improving the performance of the existing state-of-the-art models has become increasingly challenging. For this reason, the paradigm for neural architecture design is shifting from being expert-driven to almost fully automated. An emerging body of research related to such machine-aided design is called a Neural Architecture Search (NAS). This paper reviews the recent works on NAS and highlights several crucial concepts and problems of this field.},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{FAN2022107968,

title = {Self-attention neural architecture search for semantic image segmentation},

author = {Zhenkun Fan and Guosheng Hu and Xin Sun and Gaige Wang and Junyu Dong and Chi Su},

url = {https://www.sciencedirect.com/science/article/pii/S0950705121010947},

doi = {https://doi.org/10.1016/j.knosys.2021.107968},

issn = {0950-7051},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {Knowledge-Based Systems},

volume = {239},

pages = {107968},

abstract = {Self-attention can capture long-distance dependencies and is widely used in semantic segmentation. Existing methods mainly use two kinds of self-attentions, i.e., spatial attention and channel attention, which can capture the relations in HW dimension (image plane, height and width) and C dimension (channels), respectively. Very little research investigates self-attention along other dimensions, which can potentially improve the segmentation performance. In this work, we investigate the self-attentions along all the possible dimensions H,W,C,HW,HC,CW,HWC. Then we explore the aggregation of all the possible self-attentions. We apply the neural architecture search (NAS) technique to achieve optimal aggregation. Specifically, we carefully design (1) the search space and (2) the optimization method. For (1), we introduce a building block, a basic self-attention search unit (BSU), which can model self-attentions along all the dimensions. And the search space contains within-BSU and cross-BSU operations. In addition, we propose an attention-map splitting method, which can reduce the computations by 1/3. For (2), we apply an efficient differentiable optimization method to search the optimal aggregation. We conduct extensive experiments on Cityscapes and ADE20K datasets. The results show the effectiveness of the proposed method, and we achieve very competitive performance against state-of-the-art methods.},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{LIU2022102316,

title = {NAS-optimized topology-preserving transfer learning for differentiating cortical folding patterns},

author = {Shengfeng Liu and Fangfei Ge and Lin Zhao and Tianfu Wang and Dong Ni and Tianming Liu},

url = {https://www.sciencedirect.com/science/article/pii/S1361841521003613},

doi = {https://doi.org/10.1016/j.media.2021.102316},

issn = {1361-8415},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {Medical Image Analysis},

volume = {77},

pages = {102316},

abstract = {Increasing evidence suggests that cortical folding patterns of human cerebral cortex manifest overt structural and functional differences. However, for interpretability, few studies leverage advanced techniques (e.g., deep learning) to investigate the difference among cortical folds, resulting in more differences yet to be extensively explored. To this end, we proposed an effective topology-preserving transfer learning framework to differentiate cortical fMRI time series extracted from cortical folds. Our framework consists of three main parts: (1) Neural architecture search (NAS), which is used to devise a well-performing network structure based on an initialized hand-designed super-graph in an image dataset; (2) Topology-preserving transfer, which takes the model searched by NAS as the source network, keeping the topological connectivity in the network unchanged, while transforming all 2D operations including convolution and pooling into 1D, therefore resulting in a topology-preserving network, named TPNAS-Net; (3) Classification and correlation analysis, which involves using the TPNAS-Net to classify 1D cortical fMRI time series for each individual brain, and performing a group difference analysis between autism spectrum disorder (ASD) and healthy control (HC) and correlation analysis with clinical information (i.e., age). Extensive experiments on two ASD datasets obtain consistent results, demonstrating that the TPNAS-Net not only discriminates cortical folding patterns at high classification accuracy, but also captures subtle differences between ASD and HC (p-value = 0.042). In addition, we discover that there is a positive correlation between the classification accuracy and age in ASD (r = 0.39},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{SIVANGI202263,

title = {NoAS-DS: Neural optimal architecture search for detection of diverse DNA signals},

author = {Kaushik Bhargav Sivangi and Chandra Mohan Dasari and Santhosh Amilpur and Raju Bhukya},

url = {https://www.sciencedirect.com/science/article/pii/S0893608021004822},

doi = {https://doi.org/10.1016/j.neunet.2021.12.009},

issn = {0893-6080},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {Neural Networks},

volume = {147},

pages = {63-71},

abstract = {Neural network architectures are high-performing variable models that can solve many learning tasks. Designing architectures manually require substantial time and also prior knowledge and expertise to develop a high-accuracy model. Most of the architecture search methods are developed over the task of image classification resulting in the building of complex architectures intended for large data inputs such as images. Motivated by the applications of DNA computing in Neural Architecture Search (NAS), we propose NoAS-DS which is specifically built for the architecture search of sequence-based classification tasks. Furthermore, NoAS-DS is applied to the task of predicting binding sites. Unlike other methods that implement only Convolution layers, NoAS-DS, specifically combines Convolution and LSTM layers that helps in the process of automatic architecture building. This hybrid approach helped in achieving high accuracy results on TFBS and RBP datasets which outperformed other models in TF-DNA binding prediction tasks. The best architectures generated by the proposed model can be applied to other DNA datasets of similar nature using transfer learning technique that demonstrates its generalization capability. This greatly reduces the effort required to build new architectures for other prediction tasks.},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{ZHANG2022107984,

title = {On streaming disaster damage assessment in social sensing: A crowd-driven dynamic neural architecture searching approach},

author = {Yang Zhang and Ruohan Zong and Ziyi Kou and Lanyu Shang and Dong Wang},

url = {https://www.sciencedirect.com/science/article/pii/S0950705121011023},

doi = {https://doi.org/10.1016/j.knosys.2021.107984},

issn = {0950-7051},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {Knowledge-Based Systems},

volume = {239},

pages = {107984},

abstract = {Motivated by the recent advances in Internet and communication techniques and the proliferation of online social media, social sensing has emerged as a new sensing paradigm to obtain timely observations of the physical world from “human sensors”. In this study, we focus on an emerging application in social sensing – streaming disaster damage assessment (DDA), which aims to automatically assess the damage severity of affected areas in a disaster event on the fly by leveraging the streaming imagery data about the disaster on social media. In particular, we study a dynamic optimal neural architecture searching (NAS) problem in streaming DDA applications. Our goal is to dynamically determine the optimal neural network architecture that accurately estimates the damage severity for each newly arrived image in the stream by leveraging human intelligence from the crowdsourcing systems. The present study is motivated by the observation that the neural network architectures in current DDA solutions are mainly designed by artificial intelligence (AI) experts, which often leads to non-negligible costs and errors given the dynamic nature of the streaming DDA applications and the lack of real-time annotations of the massive social media data inputs. Two critical technical challenges exist in solving our problem: (i) it is non-trivial to dynamically identify the optimal neural network architecture for each image on the fly without knowing its ground-truth label a priori; (ii) it is challenging to effectively leverage the imperfect crowd intelligence to correctly identify the optimal neural network architecture for each image. To address the above challenges, we developed CD-NAS, a dynamic crowd-AI collaborative NAS framework that carefully explores the human intelligence from crowdsourcing systems to solve the dynamic optimal NAS problem and optimize the performance of streaming DDA applications. The evaluation results from a real-world streaming DDA application show that CD-NAS consistently outperforms the state-of-the-art AI and NAS baselines by achieving the highest disaster damage assessment accuracy while maintaining the lowest computational cost.},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{CHEN202238,

title = {PT-NAS: Designing efficient keypoint-based object detectors for desktop CPU platforms},

author = {Dong Chen and Hao Shen and Yuchen Shen},

url = {https://www.sciencedirect.com/science/article/pii/S0925231221019238},

doi = {https://doi.org/10.1016/j.neucom.2021.12.067},

issn = {0925-2312},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {Neurocomputing},

volume = {476},

pages = {38-52},

abstract = {Recently, keypoint-based object detectors have attracted widespread attention, due to their novel structure and excellent performance. However, in terms of their design, there are still two limitations: 1) Most keypoint-based methods are designed for GPU platforms, which makes them inefficient on desktop CPU platforms. 2) Existing works still rely heavily on manual design and prior knowledge. To this end, this work aims to offer a practical solution for designing CPU-efficient key-point detectors. First, we present a set of practical design guidelines by comparing different detection architectures. Following the proposed guidelines, we further develop a progressive three-phase network architecture search (PT-NAS) to achieve the automated design of detection architectures. Benefiting from our hierarchical search space and novel search pipeline, our PT-NAS not only achieves higher search efficiency, but also satisfies the practicality of CPU platforms. On the MS-COCO benchmark, we utilize our PT-NAS to generate several key-point detectors for fast inference on desktop CPUs. Finally, comprehensive comparison experiments prove that the proposed PT-NAS can produce new state-of-the-art keypoint-based detectors for CPU platforms.},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{rs14010141,

title = {RS-DARTS: A Convolutional Neural Architecture Search for Remote Sensing Image Scene Classification},

author = {Zhen Zhang and Shanghao Liu and Yang Zhang and Wenbo Chen},

url = {https://www.mdpi.com/2072-4292/14/1/141},

doi = {10.3390/rs14010141},

issn = {2072-4292},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {Remote Sensing},

volume = {14},

number = {1},

abstract = {Due to the superiority of convolutional neural networks, many deep learning methods have been used in image classification. The enormous difference between natural images and remote sensing images makes it difficult to directly utilize or modify existing CNN models for remote sensing scene classification tasks. In this article, a new paradigm is proposed that can automatically design a suitable CNN architecture for scene classification. A more efficient search framework, RS-DARTS, is adopted to find the optimal network architecture. This framework has two phases. In the search phase, some new strategies are presented, making the calculation process smoother, and better distinguishing the optimal and other operations. In addition, we added noise to suppress skip connections in order to close the gap between trained and validation processing and ensure classification accuracy. Moreover, a small part of the neural network is sampled to reduce the redundancy in exploring the network space and speed up the search processing. In the evaluation phase, the optimal cell architecture is stacked to construct the final network. Extensive experiments demonstrated the validity of the search strategy and the impressive classification performance of RS-DARTS on four public benchmark datasets. The proposed method showed more effectiveness than the manually designed CNN model and other methods of neural architecture search. Especially, in terms of search cost, RS-DARTS consumed less time than other NAS methods.},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{jlpea12010002,

title = {A Framework for Ultra Low-Power Hardware Accelerators Using NNs for Embedded Time Series Classification},

author = {Daniel Reiser and Peter Reichel and Stefan Pechmann and Maen Mallah and Maximilian Oppelt and Amelie Hagelauer and Marco Breiling and Dietmar Fey and Marc Reichenbach},

url = {https://www.mdpi.com/2079-9268/12/1/2},

doi = {10.3390/jlpea12010002},

issn = {2079-9268},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {Journal of Low Power Electronics and Applications},

volume = {12},

number = {1},

abstract = {In embedded applications that use neural networks (NNs) for classification tasks, it is important to not only minimize the power consumption of the NN calculation, but of the whole system. Optimization approaches for individual parts exist, such as quantization of the NN or analog calculation of arithmetic operations. However, there is no holistic approach for a complete embedded system design that is generic enough in the design process to be used for different applications, but specific in the hardware implementation to waste no energy for a given application. Therefore, we present a novel framework that allows an end-to-end ASIC implementation of a low-power hardware for time series classification using NNs. This includes a neural architecture search (NAS), which optimizes the NN configuration for accuracy and energy efficiency at the same time. This optimization targets a custom designed hardware architecture that is derived from the key properties of time series classification tasks. Additionally, a hardware generation tool is used that creates a complete system from the definition of the NN. This system uses local multi-level RRAM memory as weight and bias storage to avoid external memory access. Exploiting the non-volatility of these devices, such a system can use a power-down mode to save significant energy during the data acquisition process. Detection of atrial fibrillation (AFib) in electrocardiogram (ECG) data is used as an example for evaluation of the framework. It is shown that a reduction of more than 95% of the energy consumption compared to state-of-the-art solutions is achieved.},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{DING2022,

title = {NAP: Neural Architecture search with Pruning},

author = {Yadong Ding and Yu Wu and Chengyue Huang and Siliang Tang and Fei Wu and Yi Yang and Wenwu Zhu and Yueting Zhuang},

url = {https://www.sciencedirect.com/science/article/pii/S0925231221018361},

doi = {https://doi.org/10.1016/j.neucom.2021.12.002},

issn = {0925-2312},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {Neurocomputing},

abstract = {There has been continuously increasing attention attracted by Neural Architecture Search (NAS). Due to its computational efficiency, gradient-based NAS methods like DARTS have become the most popular framework for NAS tasks. Nevertheless, as the search iterates, the derived model in previous NAS frameworks becomes dominated by skip-connects, causing the performance downfall. In this work, we present a novel approach to alleviate this issue, named Neural Architecture search with Pruning (NAP). Unlike prior differentiable architecture search works, our approach draws the idea from network pruning. We first train an over-parameterized network, including all candidate operations. Then we propose a criterion to prune the network. Based on a newly designed relaxation of architecture representation, NAP can derive the most potent model by removing trivial and redundant edges from the whole network topology. Experiments show the effectiveness of our proposed approach. Specifically, the model searched by NAP achieves state-of-the-art performances (2.48% test error) on CIFAR-10. We transfer the model to ImageNet and obtains a 25.1% test error with only 5.0M parameters, which is on par with modern NAS methods.},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{RORABAUGH2022107780,

title = {High frequency accuracy and loss data of random neural networks trained on image datasets},

author = {Ariel Keller Rorabaugh and Silvina Caíno-Lores and Travis Johnston and Michela Taufer},

url = {https://www.sciencedirect.com/science/article/pii/S2352340921010544},

doi = {https://doi.org/10.1016/j.dib.2021.107780},

issn = {2352-3409},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {Data in Brief},

volume = {40},

pages = {107780},

abstract = {Neural Networks (NNs) are increasingly used across scientific domains to extract knowledge from experimental or computational data. An NN is composed of natural or artificial neurons that serve as simple processing units and are interconnected into a model architecture; it acquires knowledge from the environment through a learning process and stores this knowledge in its connections. The learning process is conducted by training. During NN training, the learning process can be tracked by periodically validating the NN and calculating its fitness. The resulting sequence of fitness values (i.e., validation accuracy or validation loss) is called the NN learning curve. The development of tools for NN design requires knowledge of diverse NNs and their complete learning curves. Generally, only final fully-trained fitness values for highly accurate NNs are made available to the community, hampering efforts to develop tools for NN design and leaving unaddressed aspects such as explaining the generation of an NN and reproducing its learning process. Our dataset fills this gap by fully recording the structure, metadata, and complete learning curves for a wide variety of random NNs throughout their training. Our dataset captures the lifespan of 6000 NNs throughout generation, training, and validation stages. It consists of a suite of 6000 tables, each table representing the lifespan of one NN. We generate each NN with randomized parameter values and train it for 40 epochs on one of three diverse image datasets (i.e., CIFAR-100, FashionMNIST, SVHN). We calculate and record each NN’s fitness with high frequency—every half epoch—to capture the evolution of the training and validation process. As a result, for each NN, we record the generated parameter values describing the structure of that NN, the image dataset on which the NN trained, and all loss and accuracy values for the NN every half epoch. We put our dataset to the service of researchers studying NN performance and its evolution throughout training and validation. Statistical methods can be applied to our dataset to analyze the shape of learning curves in diverse NNs, and the relationship between an NN’s structure and its fitness. Additionally, the structural data and metadata that we record enable the reconstruction and reproducibility of the associated NN.},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{9674227,

title = {Building High-throughput Neural Architecture Search Workflows via a Decoupled Fitness Prediction Engine},

author = {Ariel Keller Rorabaugh and Silvina Caino-Lores and Travis Johnston and Michela Taufer},

url = {https://ieeexplore.ieee.org/abstract/document/9674227},

doi = {10.1109/TPDS.2022.3140681},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {IEEE Transactions on Parallel and Distributed Systems},

pages = {1-1},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{9673794,

title = {Automated model generation for machinery fault diagnosis based on reinforcement learning and neural architecture search},

author = {Jian Zhou and Lianyu Zheng and Yiwei Wang and Cheng Wang and Robert X. Gao},

url = {https://ieeexplore.ieee.org/abstract/document/9673794},

doi = {10.1109/TIM.2022.3141166},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {IEEE Transactions on Instrumentation and Measurement},

pages = {1-1},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{9672175,

title = {Evolutionary Search for Complete Neural Network Architectures with Partial Weight Sharing},

author = {Haoyu Zhang and Yaochu Jin and Yaochu Jin and Kuangrong Hao},

url = {https://ieeexplore.ieee.org/abstract/document/9672175},

doi = {10.1109/TEVC.2022.3140855},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {IEEE Transactions on Evolutionary Computation},

pages = {1-1},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{YIPENG2022108819,

title = {Determination of wheat kernels damaged by Fusarium head blight using monochromatic images of effective wavelengths from hyperspectral imaging coupled with an architecture self-search deep network},

author = {Lv Yipeng and Lv Wenbing and Han Kaixuan and Tao Wentao and Zheng Ling and Weng Shizhuang and Huang Lingsheng},

url = {https://www.sciencedirect.com/science/article/pii/S0956713522000123},

doi = {https://doi.org/10.1016/j.foodcont.2022.108819},

issn = {0956-7135},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {Food Control},

pages = {108819},

abstract = {Wheat kernels damaged by Fusarium head blight (FHB) lose moisture, protein and starch and carry dangerous toxins. Classification of degree of damage can be helpful to customise the use of wheat kernels, reduce financial losses and ensure grain safety. In this study, hyperspectral imaging (HSI) and deep learning network were explored to determine sound, mildly, moderately and severely damaged wheat kernels. Effective wavelengths (EWs) were selected from the reflectance spectroscopy of HSI images by ReliefF, uninformative variable elimination, random frog and shuffled frog leaping algorithm, and the monochromatic images of different combinations of EWs were adopted to develop the classification models combined with an architecture self-search deep network (ASSDN). ASSDN and the images composed of 941, 876 and 732 nm achieved the best determination with average accuracy of 100% and 98.31% in the training and prediction sets, respectively, outperforming other images or methods. And the average area under the curve of 0.9985 indicated its excellent robustness. Using images of sporadic wavelengths, the computation and operation complexity were evidently reduced, and a simple and custom-built instrument can be easily designed for practical recognition of FHB-damaged wheat kernels. Meanwhile, ASSDN can generate and optimise the high-performance classification network by itself, which is user–friendly and largely expands application potential of deep network.},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{TAN2022103495,

title = {Neural architecture search for real-time quality assessment of wearable multi-lead ECG on mobile devices},

author = {Huixin Tan and Jiewei Lai and Yunbi Liu and Yuzhang Song and Jinliang Wang and Mingyang Chen and Yong Yan and Liming Zhong and Qianjin Feng and Wei Yang},

url = {https://www.sciencedirect.com/science/article/pii/S1746809422000179},

doi = {https://doi.org/10.1016/j.bspc.2022.103495},

issn = {1746-8094},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {Biomedical Signal Processing and Control},

volume = {74},

pages = {103495},

abstract = {Users need to pay for the medical services of the wearable ECG signals. The complicated interferences of wearable ECG may cause invalid medical diagnosis. To save medical expenses and reduce the workload of cardiologists, unacceptable wearable ECG should be filtered out on mobile devices in real time. Therefore, a robust and lightweight quality assessment model is essential. Existing convolutional neural network models are more robust than hand-crafted features based models, but most are too cumbersome to run on mobile devices. In this study, we utilize neural architecture search algorithm to search a network with a few parameters and low latency automatically for real-time quality assessment of wearable ECG data. To reach this goal, we construct a searching space with lightweight network blocks and add the hardware latency loss. We also propose a novel and effective strategy, any-lead, to assess the quality of all leads using a single model. We evaluated our method on a large-scale (10,709 signals) wearable 12-lead ECG dataset and a public dataset named Physionet Cinc Challenge 2011. The parameters and FLOPs of the architecture we searched were about 66.76 K and 36.44 M. Our model achieved excellent performance on the aforementioned datasets, with AUC of 98.32% and 97.64%, F1 scores of 94.36% and 93.52%, MCC of 84.74% and 83.22%, respectively, and with the inference time on an Android emulator of about 78 ms. Extensive experimental results demonstrate the effectiveness of our method in assessing the quality of all leads of wearable ECG data on mobile devices in real time.},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{9684972,

title = {Federated Neural Architecture Search for Medical Data Security},

author = {Xin Liu and Jianwei Zhao and Jie Li and Bin Cao and Zhihan Lv},

url = {https://ieeexplore.ieee.org/abstract/document/9684972},

doi = {10.1109/TII.2022.3144016},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {IEEE Transactions on Industrial Informatics},

pages = {1-1},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{GUO2022108448,

title = {Differentiable neural architecture learning for efficient neural networks},

author = {Qingbei Guo and Xiao-Jun Wu and Josef Kittler and Zhiquan Feng},

url = {https://www.sciencedirect.com/science/article/pii/S0031320321006245},

doi = {https://doi.org/10.1016/j.patcog.2021.108448},

issn = {0031-3203},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {Pattern Recognition},

volume = {126},

pages = {108448},

abstract = {Efficient neural networks has received ever-increasing attention with the evolution of convolutional neural networks (CNNs), especially involving their deployment on embedded and mobile platforms. One of the biggest problems to obtaining such efficient neural networks is efficiency, even recent differentiable neural architecture search (DNAS) requires to sample a small number of candidate neural architectures for the selection of the optimal neural architecture. To address this computational efficiency issue, we introduce a novel architecture parameterization based on scaled sigmoid function, and propose a general Differentiable Neural Architecture Learning (DNAL) method to obtain efficient neural networks without the need to evaluate candidate neural networks. Specifically, for stochastic supernets as well as conventional CNNs, we build a new channel-wise module layer with the architecture components controlled by a scaled sigmoid function. We train these neural network models from scratch. The network optimization is decoupled into the weight optimization and the architecture optimization, which avoids the interaction between the two types of parameters and alleviates the vanishing gradient problem. We address the non-convex optimization problem of efficient neural networks by the continuous scaled sigmoid method instead of the common softmax method. Extensive experiments demonstrate our DNAL method delivers superior performance in terms of efficiency, and adapts to conventional CNNs (e.g., VGG16 and ResNet50), lightweight CNNs (e.g., MobileNetV2) and stochastic supernets (e.g., ProxylessNAS). The optimal neural networks learned by DNAL surpass those produced by the state-of-the-art methods on the benchmark CIFAR-10 and ImageNet-1K dataset in accuracy, model size and computational complexity. Our source code is available at https://github.com/QingbeiGuo/DNAL.git.},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{9686068,

title = {Evolutionary Architectural Search for Generative Adversarial Networks},

author = {Qiuzhen Lin and Zhixiong Fang and Yi Chen and Kay Chen Tan and Yun Li},

url = {https://ieeexplore.ieee.org/abstract/document/9686068},

doi = {10.1109/TETCI.2021.3137377},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {IEEE Transactions on Emerging Topics in Computational Intelligence},

pages = {1-12},

keywords = {},

pubstate = {published},

tppubtype = {article}

}