A Comprehensive Survey of Neural Architecture Search

Neural Architecture Search (NAS) is the process of automating the design of neural network architectures for a given task. Although NAS provides automates the process of finding suitable neural network architectures for a specific task, the existing NAS algorithms are immensely time-consuming. The main bottleneck in NAS algorithms is the training time for each architecture. This study proposes an Improved Grey Wolf Optimization based on Synaptic Saliency (IGWO-SS), which is much faster than the existing NAS algorithms and provides better final performance. The IGWO-SS algorithm skips training the less promising architectures by creating a relative rank between the architectures based on synaptic saliency. The architectures that are lower in rank are considered less promising than the architectures that are higher in rank. Since the calculation of synaptic saliency is a very fast process, a significant amount of time is saved by skipping training of less promising architectures. We performed extensive experiments to determine the efficacy of synaptic saliency in improving NAS. Our experimental results suggest that the synaptic saliency of an untrained neural network positively correlates with its final accuracy. Hence, it can be used to identify untrained promising neural networks. The experimental results suggest that the IGWO-SS algorithm is almost 10<i>x</i> faster and achieves better final performance than five other bio-inspired algorithms. The IGWO-SS algorithm achieves higher mean accuracy than state-of-the-art NAS algorithms, including - REA, RS, RL, BOHB, DARTSV1, DARTSV2, GDAS, SETN, and ENAS. We hope that our work will make NAS more accessible and useful to researchers by reducing the time and resources required to perform NAS.

With the booming development of Internet of things (IoT) devices and machine learning (ML) technique, edge machine learning is emerging to process the enormous sampled data for realizing intelligent applications at the network edge. With limited edge resources, a well-structured neural network and numerous training data are the two main factors that affect the performance of edge machine learning. In this paper, we cooperatively optimize the data collection and the neural architecture to minimize the energy consumption of devices and the error on a specific task. We derive the Rademacher complexity bounds theoretically to evaluate the generalization error of the neural architectures in the search space and then formulate the optimization problem accordingly. Then we develop a scheme to solve the problem that dynamically performs the data collection based on policy gradient reinforcement learning and the parameter-sharing neural architecture search (NAS) algorithm. By this way, the transmission power of each device can be adjusted based on the data quality assessed by the NAS result in each round to effectively collect data. And with the growing high-quality data, the NAS algorithm can gradually find the optimal architecture for the task. Experimental results show that the neural architectures found by the proposed algorithm outperform the existing architectures while saving energy in the device.

Neural Architecture Search (NAS) is the game changer in designing robust neural architectures. Architectures designed by NAS outperform or compete with the best manual network designs in terms of accuracy, size, memory footprint and FLOPs. That said, previous studies focus on developing NAS algorithms for clean high quality data, a restrictive and somewhat unrealistic assumption. In this paper, focusing on the differentiable NAS algorithms, we show that vanilla NAS algorithms suffer from a performance loss if class labels are noisy. To combat this issue, we make use of the principle of information bottleneck as a regularizer. This leads us to develop a noise injecting operation that is included during the learning process, preventing the network from learning from noisy samples. Our empirical evaluations show that the noise injecting operation does not degrade the performance of the NAS algorithm if the data is indeed clean. In contrast, if the data is noisy, the architecture learned by our algorithm comfortably outperforms algorithms specifically equipped with sophisticated mechanisms to learn in the presence of label noise. In contrast to many algorithms designed to work in the presence of noisy labels, prior knowledge about the properties of the noise and its characteristics are not required for our algorithm.

A neural network model needs to be manually designed for ancient mural dynasty recognition, and this paper proposes an ancient mural dynasty recognition algorithm that is based on a neural architecture search (NAS). First, the structural edge information of mural images is extracted for use by the neural network model in recognizing mural missions. Second, an NAS algorithm that is based on contrast selection (CS) simplifies the architecture search to an incremental CS and then searches for the optimal network architecture on the mural dataset. Finally, the identified optimal network architecture is used for training and testing to complete the mural dynasty recognition task. The results show that the top accuracy of the proposed method on the mural dataset is 88.10%, the recall rate is 87.52%, and the precision rate is 87.69%. Each evaluation index used by the neural network model is superior to that of classical network models such as AlexNet and ResNet-50. Compared with NAS methods such as ASNG and MIGO, the accuracy of mural dynasty recognition is higher by an average of 4.27% when using the proposed method. The proposed method is verified on CIFAR-10, CIFAR-100, ImageNet16-120 and other datasets and achieves a good recognition accuracy in the NAS-bench-201 search space, which averages 93.26%, 70.73% and 45.34%, respectively, on the abovementioned datasets.

Graph neural networks (GNNs) are popularly used to analyze non-Euclidean graph data. Despite their successes, the design of graph neural networks requires heavy manual work and rich domain knowledge. Recently, neural architecture search algorithms are widely used to automatically design neural architectures for CNNs and RNNs. Inspired by the success of neural architecture search algorithms, we present a graph neural architecture search algorithm GraphNAS that enables automatic design of the best graph neural architecture based on reinforcement learning. Specifically, GraphNAS uses a recurrent network as the controller to generate variable-length strings that describe the architectures of graph neural networks, and trains the recurrent network with policy gradient to maximize the expected accuracy of the generated architectures on a validation data set. Moreover, based on GraphNAS, we design a new GraphNAS++ model using distributed neural architecture search. Compared with GraphNAS that generates and evaluates only one candidate architecture at each iteration, GraphNAS++ generates a mini-batch of candidate architectures and evaluates them in a distributed computing environment until convergence. Experiments on real-world datasets demonstrate that GraphNAS can design a novel network architecture that rivals the best human-invented architecture. Moreover, GraphNAS++ can speed up the design process at least five times by using the distributed training framework with GPUs.

Designing advanced neural architectures to tackle specific tasks involves weeks or even months of intensive investigation by experts with rich domain knowledge. In recent years, neural architecture search (NAS) has attracted the interest of many researchers due to its ability to automatically design efficient neural architectures. Among different search strategies, evolutionary algorithms have achieved significant successes as derivative-free optimization algorithms. However, the tremendous computational resource consumption of the evolutionary neural architecture search dramatically restricts its application. In this paper, we explore how fitness approximation-based evolutionary algorithms can be applied to neural architecture search and propose NAS-EA-FA to accelerate the search process. We further exploit data augmentation and diversity of neural architectures to enhance the algorithm, and present NAS-EA-FA V2. Experiments show that NAS-EA-FA V2 is at least five times faster than other state-of-the-art neural architecture search algorithms like regularized evolution and iterative neural predictor on NASBench-101, and it is also the most effective and stable algorithm on NASBench-201. All the code used in this paper is available at https://github.com/fzjcdt/NAS-EA-FA.

Deep Neural Networks are often vulnerable to adversarial attacks. Neural Architecture Search (NAS), one of the tools for developing novel deep neural architectures, demonstrates superior performance in prediction accuracy in various machine learning applications. However, the performance of a neural architecture discovered by NAS against adversarial attacks has not been sufficiently studied, especially under the regime of knowledge distillation. Given the presence of a robust teacher, we investigate if NAS would produce a robust neural architecture by inheriting robustness from the teacher. In this paper, we propose Robust Neural Architecture Search by Cross-Layer knowledge distillation (RNAS-CL), a novel NAS algorithm that improves the robustness of NAS by learning from a robust teacher through cross-layer knowledge distillation. Unlike previous knowledge distillation methods that encourage close student-teacher output only in the last layer, RNAS-CL automatically searches for the best teacher layer to supervise each student layer. Experimental results demonstrate the effectiveness of RNAS-CL and show that RNAS-CL produces compact and adversarially robust neural architectures. Our results point to new approaches for finding compact and robust neural architecture for many applications. The code of RNAS-CL is available at https://github.com/Statistical-Deep-Learning/RNAS-CL.

Genetic Neural Architecture Search for automatic assessment of human sperm images

The emergence of neural architecture search (NAS) algorithms has removed the constraints on manually designed neural network architectures, so that neural network development no longer requires extensive professional knowledge, trial and error. However, the extremely high computational cost limits the development of NAS algorithms. In this article, in order to reduce computational costs and to improve the efficiency and effectiveness of evolutionary NAS (ENAS) is investigated. In this article, we present a fast ENAS framework for multiscale convolutional networks based on evolutionary knowledge transfer search (EKTS). This framework is novel, in that it combines global optimization methods with local optimization methods for search, and searches a multiscale network architecture. In this article, evolutionary computation is used as a global optimization algorithm with high robustness and wide applicability for searching neural architectures. At the same time, for fast search, we combine knowledge transfer and local fast learning to improve the search speed. In addition, we explore a multiscale gray-box structure. This gray box structure combines the Bandelet transform with convolution to improve network approximation, learning, and generalization. Finally, we compare the architectures with more than 40 different neural architectures, and the results confirmed its effectiveness.

Motivated by the observation that most neural architecture search (NAS) methods are time consuming because a "training process" is required to evaluate each searched neural architecture, this article presents an efficient NAS algorithm based on a promising metaheuristic algorithm named search economics (SE) and a new training-free estimator to evaluate the searched neural architectures for not only obtaining a good neural architecture but also accelerating the computation time. The basic idea of the proposed NAS algorithm is to use the so-called expected value of each region in the search space to guide the search so that it will focus on searching high potential regions instead of solutions with high objective values in particular regions. To evaluate the performance of the proposed algorithm, we compare it with state-of-the-art nontraining-free and training-free NAS methods. Experimental results show that the proposed algorithm is capable of finding a result that is similar to or better than those found by most nontraining-free NAS algorithms compared in this study but taking only a tiny portion of the computation time.

ETNAS: An energy consumption task-driven neural architecture search

Convolution neural networks (CNN) have emerged as a prevailing paradigm for micro-expression recognition (MER) yet, it is inefficient and time-intensive to design optimal CNN-based MER models manually. In recent times, the neural architecture search (NAS) has garnered attention due to its automatic CNN architecture searching ability. However, the performance of NAS in MER is limited by challenges such as rapid duration, subtle intensity, and a mismatch between architecture and cell-level search. The existing search space, which stacks twelve cells with three transition paths (downsample, upsample, and same resolution), creates deep networks that may diminish minute spatio-temporal features due to progressive convolution and pooling. Therefore, motivated by these factors, in this paper, we introduces a novel approach, the Micro-Expression Feature Adaptive Neural Architecture Search (ME-NAS), to analyze true human emotions through micro-expression recognition. While Neural Architecture Search (NAS) has gained attention for its automatic Convolutional Neural Network (CNN) architecture search ability, its application in Micro-Expression Recognition (MER) faces challenges due to ingrained challenges (rapid duration, subtle and low intensity) and the discrepancy between architecture and cell-level search. The existing NAS architecture search space is designed by stacking twelve cells with three transition paths (downsample, upsample, and same resolution), resulting in a deep network. Such deep networks may diminish minute spatiotemporal features due to the progressive convolution and pooling operations. Motivated by these factors, we designed a new NAS algorithm: ME-NAS. The ME-NAS comprises expressive feature exploration (EXPERT) in architecture search along with refined and complementary feature derivative (ReCODE) operations in cell-level search. The EXPERT aims to trace the optimal paths instead of covering all possible paths between cells. The ReCODE operations capture micro-level variations from spatial and temporal domains by introducing 24 3D-convolution operations. The proposed ReCODE and EXPERT search space jointly lead to the search for a robust and shallow CNN architecture for micro expressions (MEs). The proposed ME-NAS is evaluated on six datasets: CASME-I, CASME-II, CAS(ME) 2 , SAMM, SMIC, and MEGC-19 composite, with two evaluation strategies: LOSO, and cross-domain, respectively. The experimental results manifest that the proposed ME-NAS outperformed the state-of-the-art approaches on both evaluation strategies.

Despite the increasing interest in neural architecture search (NAS), the significant computational cost of NAS is a hindrance to researchers. Hence, we propose to reduce the cost of NAS using proxy data, i.e., a representative subset of the target data, without sacrificing search performance. Even though data selection has been used across various fields, our evaluation of existing selection methods for NAS algorithms offered by NAS-Bench-1shot1 reveals that they are not always appropriate for NAS and a new selection method is necessary. By analyzing proxy data constructed using various selection methods through data entropy, we propose a novel proxy data selection method tailored for NAS. To empirically demonstrate the effectiveness, we conduct thorough experiments across diverse datasets, search spaces, and NAS algorithms. Consequently, NAS algorithms with the proposed selection discover architectures that are competitive with those obtained using the entire dataset. It significantly reduces the search cost: executing DARTS with the proposed selection requires only 40 minutes on CIFAR-10 and 7.5 hours on ImageNet with a single GPU. Additionally, when the architecture searched on ImageNet using the proposed selection is inversely transferred to CIFAR-10, a state-of-the-art test error of 2.4% is yielded. Our code is available at https://github.com/nabk89/NAS-with-Proxy-data.

Neural architecture search (NAS) is an automated technique to design optimal neural network architectures for a specific workload. Conventionally, evaluating candidate networks in NAS involves extensive training, which requires significant time and computational resources. To address this, training-free NAS has been proposed to expedite network evaluation with minimal search time. However, state-of-the-art training-free NAS algorithms struggle to precisely distinguish well-performing networks from poorly performing networks, resulting in inaccurate performance predictions and consequently suboptimal top-one network accuracy. Moreover, they are less effective in activation function exploration. To tackle the challenges, this article proposes RBFleX-NAS, a novel training-free NAS framework that accounts for both activation outputs and input features of the last layer with a radial basis function (RBF) kernel. We also present a detection algorithm to identify optimal hyperparameters using the obtained activation outputs and input feature maps. We verify the efficacy of RBFleX-NAS over a variety of NAS benchmarks. RBFleX-NAS significantly outperforms state-of-the-art training-free NAS methods in terms of top-one accuracy, achieving this with short search time in NAS-Bench-201 and NAS-Bench-SSS. In addition, it demonstrates a higher Kendall correlation compared to layer-based training-free NAS algorithms. Furthermore, we propose the neural network activation function benchmark (NAFBee), a new activation design space that extends the activation type to encompass various commonly used functions. In this extended design space, RBFleX-NAS demonstrates its superiority by accurately identifying the best-performing network during activation function search, providing a significant advantage over other NAS algorithms.

From the perspective of data stream, neural architecture search (NAS) can be formulated as a graph optimization problem. However, many state-of-the-art black-box optimization algorithms, such as Bayesian optimization and simulated annealing, operate in continuous space primarily, which does not match the NAS optimization due to the discreteness of graph structures. To tackle this problem, the latent space Bayesian optimization NAS (LSBO-NAS) algorithm is developed in this paper. In LSBO-NAS, the neural architectures are represented as sequences, and a variational auto-encoder (VAE) is trained to convert the discrete search space of NAS into a continuous latent space by learning the continuous representation of neural architectures. Hereafter, a Bayesian optimization (BO) algorithm, i.e., the tree-structure parzen estimator (TPE) algorithm, is developed to obtain admirable neural architectures. The optimization loop of LSBO-NAS consists of two stages. In the first stage, the BO algorithm generates a preferable architecture representation according to its search strategy. In the second stage, the decoder of VAE decodes the representation into a discrete neural architecture, whose performance evaluation is regarded as the feedback signal for the BO algorithm. The effectiveness of the developed LSBO-NAS is demonstrated on the NAS-Bench-301 benchmark, where the LSBO-NAS achieves a better performance than several NAS baselines.