Adversarial Learning Strategy Research Articles

A Spatial and Adversarial Representation Learning Approach for Land Use Classification with POIs

Points-of-interests (POIs) have been proven to be indicative for sensing urban land use in numerous studies. However, recent progress mainly relies on spatial co-occurrence patterns among POI categories, which falls short in utilizing the rich semantic information embodied in POI hierarchical categories and in sensing the spatial distribution patterns of POIs at an individual zonal scale. In this context, we present a spatial and adversarial representation learning approach (SARL) for predicting land use of urban zones with POIs. SARL deeply mines the information from POIs from both spatial and categorical perspectives. Specifically, we first utilize a convolutional neural network to sense the spatial distribution patterns of POIs in each urban zone. We then leverage an autoencoder and an adversarial learning strategy to mine the POI categorical information in all hierarchical levels, which emphasizes the prominent and definitive POIs while preserves the overall POI hierarchical structures in each zone. Finally, we fuse these information from the two perspectives via a Wide & Deep network and carry out land use prediction with the fused embeddings. We conduct comprehensive experiments to validate the effectiveness of SARL in four European cities with real-world data. The results demonstrate that SARL substantially outperforms several competitive baselines.

Bearing performance degradation assessment using adversarial fusion convolutional autoencoder based on multi-source information

Bearing operation states will directly determine the performance of the equipment; thus, monitoring operation status and degradation indicators is the key to ensuring continuous and healthy operation of the equipment. However, most of the research uses single-source information data, which makes it difficult to model when dealing with multi-source information, complex data distribution, and noise. In this paper, a bearing performance degradation assessment method based on multi-source information is proposed to comprehensively utilize the data signals of different structures, spaces, types, and sources. First, the adversarial fusion convolutional autoencoder is constructed for obtaining the degradation index of the bearing, while the adversarial learning strategy is applied to achieve the effect of enhancing the robustness and sensitivity of the degradation indicators extracted by the network. Then the degradation index is input into the support vector data description to determine the fault anomalies of the degradation index adaptively and the fuzzy c-means algorithms to obtain the final rolling bearing performance degradation evaluation results. Through the verification results of two experiment datasets, it is found that the proposed model can achieve accurate evaluation and quantitative analysis of the performance degradation process of bearings. As a result, the entire network ensures the reconstruction accuracy of normal samples while simultaneously stretching the reconstruction error of abnormal samples to achieve accurate monitoring of degradation onset.

Disentanglement Translation Network for multimodal sentiment analysis

Obtaining an effective joint representation has always been the goal for multimodal tasks. However, distributional gap inevitably exists due to the heterogeneous nature of different modalities, which poses burden on the fusion process and the learning of multimodal representation. The imbalance of modality dominance further aggravates this problem, where inferior modalities may contain much redundancy that introduces additional variations. To address the aforementioned issues, we propose a Disentanglement Translation Network (DTN) with Slack Reconstruction to capture desirable information properties, obtain a unified feature distribution and reduce redundancy. Specifically, the encoder–decoder-based disentanglement framework is adopted to decouple the unimodal representations into modality-common and modality-specific subspaces, which explores the cross-modal commonality and diversity, respectively. In the encoding stage, to narrow down the discrepancy, a two-stage translation is devised to incorporate with the disentanglement learning framework. The first stage targets at learning modality-invariant embedding for modality-common information with adversarial learning strategy, capturing the commonality shared across modalities. The second stage considers the modality-specific information that reveals diversity. To relieve the burden of multimodal fusion, we realize Specific-Common Distribution Matching to further unify the distribution of the desirable information. As for the decoding and reconstruction stage, we propose Slack Reconstruction to seek a balance between retaining discriminative information and reducing redundancy. Although the existing commonly-used reconstruction loss with strict constraint lowers the risk of information loss, it easily leads to the preservation of information redundancy. In contrast, Slack Reconstruction imposes a more relaxed constraint so that the redundancy is not forced to be retained, and simultaneously explores the inter-sample relationships. The proposed method aids multimodal fusion by learning the exact properties and obtaining a more uniform distribution for cross-modal data, and manages to reduce information redundancy to further ensure feature effectiveness. Extensive experiments on the task of multimodal sentiment analysis indicate the effectiveness of the proposed method. The codes are available at https://github.com/zengy268/DTN.

Transfer easy to hard: Adversarial contrastive feature learning for unsupervised person re-identification

Unsupervised Person Re-Identification (Re-ID) is challenging due to the lack of ground-truth labels. Most existing methods address this problem by progressively mining high-confidence pseudo labels to guide the feature learning process. However, how to construct hard-enough samples while maintaining the fidelity of pseudo labels in these samples remains an open issue in the machine learning community. To tackle this challenge, we design a simple yet effective adversarial contrastive feature learning (ACFL) framework, which enhances the discriminative capability of features by introducing more transformed hard samples in the feature learning process. Specifically, it mainly consists of a discriminative feature learning module and a hard sample generation module. The discriminative feature learning module extracts recognizable features of unlabeled training samples to estimate the high-confidence relationship between samples. Then, the hard sample generation module utilizes these high-confidence relationships between samples to transfer all samples into the hard ones via an adversarial learning strategy. Finally, the generated hard samples are further fed into DFL to learn discriminative features for person Re-ID. Extensive experiments on Market-1501, DukeMTMC-reID, and MSMT17 datasets show that our method compares favorably with state-of-the-art methods.

UDATNN: A modeling scheme integrating unsupervised domain adversarial learning and tri-training strategy for variety recognition of maize seeds with domain shift

Constructing an accurate classification model is the key to realizing maize seed varieties identification based on optical sensing techniques. However, the optical information (features) of seeds collected by optical sensing technology is easily affected by the planting environment (e.g., origin, year), which makes the training samples used to construct the model (source domain, SD) and the samples to be recognized (target domain, TD) subject to domain shift (DS), and ultimately undermines the recognition performance of the model. In this study, a modeling scheme, called unsupervised domain adversarial tri-training of neural networks (UDATNN), is proposed for maize variety recognition in the presence of DS. Firstly, an unsupervised domain adversarial learning approach is used to map the raw features of the SD and TD into a low-dimensional feature space in order to achieve feature alignment of the SD and TD, and to improve the feature discriminability of different varieties of seeds. Subsequently, the aligned low-dimensional features are used as inputs of classifiers (random forest) and part of the target domain samples are iteratively selected and given pseudo-labels according to the tri-training strategy. Finally, these samples assigned with pseudo-labels (called updating samples) together with the training samples in SD are used to re-train the classification model constructed based on the unsupervised domain adversarial strategy, for improving recognition accuracy of seed varieties with DS scenario. Hyperspectral images of a total of 4584 maize seeds, including 7 varieties (each variety produced from two or three years) were collected to verify the performance of the proposed scheme. The recognition accuracy of the target domain reaches 91.5%, 94.1%, and 92.6% under 3 different domain shifts, which improves the recognition performance nearly by 8%–22% compared with the no-transfer model and the traditional transfer model. The model performance was further discussed by calculating precision, recall, and F1 score to achieve satisfactory results. The robustness of the model was also verified by discussing the randomness of the update samples and the effect of the number of samples in the source domain on the model performance through 100 random experiments and multiple experimental comparisons. The proposed UDATNN scheme can be used as a new framework to address the nondestructive identification of seed varieties under domain shift conditions.

Syncretic Space Learning Network for NIR-VIS Face Recognition

To overcome the technical bottleneck of face recognition in low-light scenarios, Near-InfraRed and VISible (NIR-VIS) heterogeneous face recognition is proposed for matching well-lit VIS faces with poorly lit NIR faces. Current cross-modal synthesis methods visually convert the NIR modality to the VIS modality and then perform face matching in the VIS modality. However, using a heavyweight GAN network on unpaired NIR-VIS faces may lead to high synthesis difficulty, low inference efficiency, and other problems. To alleviate the above problems, we simultaneously synthesize NIR and VIS images into modality-independent syncretic images and propose a novel syncretic space learning (SSL) model to eliminate the modal gap. First, Syncretic Modality Generator (SMG) synthesizes NIR and VIS images into syncretic images using channel-level convolution with a shallow CNN. In particular, the discriminative structural information is well preserved and the face quality can be further improved with small modal variations in a self-supervised learning manner. Second, Modality-adversarial Syncretic space Learning (MSL) projects NIR and VIS images into the syncretic space by a syncretic-modality adversarial learning strategy with syncretic pattern guided objective, so the modal gap of NIR-VIS faces can be effectively reduced. Finally, the Syncretic Distribution Consistency (SDC) constructed by NIR-syncretic, syncretic-syncretic, and VIS-syncretic consistency can enhance the intra-class compactness and learn discriminative representations. Extensive experiments on three challenging datasets demonstrate the effectiveness of the SSL method.

Study on the Flipped Classroom Teaching Model of “Business English Translation” in the Context of Big Data

Abstract As a new teaching method born in the changing times, the flipped classroom also provides a new way to reform the teaching mode of business English translation. In this paper, we propose a new neural machine translation model--TwinGAN model, which includes two generators and two discriminators and combines “similarity selection” and “strategy gradient” of reinforcement learning and adversarial learning strategy of the generative adversarial network to improve the translation quality of the model. The model combines “similarity selection”, “policy gradient,” reinforcement learning, and the adversarial learning strategy of the generative adversarial network to improve the translation quality of the model while mitigating the exposure bias problem to improve the translation performance of the model further. Finally, the learning-assisted translation system based on the TwinGAN model is combined with the deep learning-based flipped classroom teaching model and applied to business English translation teaching. In the eight-week teaching test, 38.16% of the students in the experimental class achieved excellent learning levels in the first week of the test, and 57.47% in the eighth week of the test, an increase of 18.55% compared with the first week. In contrast, 36.48% of the students in the control class reached an excellent level in the first week of testing, and 35.73% in the control class reached an excellent level in the eighth week. The flipped classroom model based on deep learning is effective in teaching effectiveness.

ADfM-Net: An Adversarial Depth-From-Motion Network Based on Cross Attention and Motion Enhanced

The temporal consistent and accurate depth estimation for consecutive images is essential for many downstream applications. However, most existing methods only infer depth from a single image, ignoring the temporal information and important depth cues from motion in the sequence. Additionally, the depths of adjacent frames are estimated separately without any constraint. In this paper, we promote the temporal consistency and accuracy of depth results from the aforementioned two aspects: multi-frame framework and consistency constraint. Firstly, a framework with cross-frame attention and motion enhancement module is proposed for better temporal consistency and depth precision. Secondly, an adversarial metric learning strategy is introduced to further constrain the consistency of adjacent depth results, without any additional computation and memory cost. The experiments on KITTI and Cityscapes datasets demonstrate the effectiveness of our framework. Furthermore, noting that the traditional metrics can not reveal the consistency of the depth results, a new temporal consistency metric is proposed, which would facilitate further research.

A histogram-driven generative adversarial network for brain MRI to CT synthesis

Magnetic Resonance Imaging (MRI) and Computed Tomography (CT) are commonly used tools for medical diagnostic assessment. Considering the ionizing radiation of CT imaging, estimating CT images from radiation-free MRI would be beneficial for medical diagnosis. Although state-of-art generative adversarial networks or end-to-end architecture models can generate realistic natural images, it is challenging to generate medical CT images with high signal-to-noise ratios and are paired with MRI. We propose a histogram-driven generative adversarial network (HisGAN) to address this issue, estimate CT images paired with MR, and develop a histogram-based dynamic scaling factor to facilitate learning different image styles. By employing an adversarial learning strategy to train the end-to-end generator, the generator better simulate the nonlinear mapping from source to target. For the generator, the proposed method applies multiple learnable parameters to adjust the overall weights of the dilated convolution layers to ensure sufficient expansive receptive fields for improved performance. Additionally, the method utilizes deep residual networks to train randomly smoothed generated images and employs adversarial loss to enhance the generation of the discriminator, achieving a balance between the generator and the discriminator. Our approach can synthesize image details at the pixel level in the target domain and has been evaluated using two datasets for MR to CT, T1 to T2, and Flair to T1ce modality synthesis tasks. The proposed method outperforms existing generative adversarial models applied to medical image synthesis.

Memory-Aided Contrastive Consensus Learning for Co-salient Object Detection

Co-salient object detection (CoSOD) aims at detecting common salient objects within a group of relevant source images. Most of the latest works employ the attention mechanism for finding common objects. To achieve accurate CoSOD results with high-quality maps and high efficiency, we propose a novel Memory-aided Contrastive Consensus Learning (MCCL) framework, which is capable of effectively detecting co-salient objects in real time (∼150 fps). To learn better group consensus, we propose the Group Consensus Aggregation Module (GCAM) to abstract the common features of each image group; meanwhile, to make the consensus representation more discriminative, we introduce the Memory-based Contrastive Module (MCM), which saves and updates the consensus of images from different groups in a queue of memories. Finally, to improve the quality and integrity of the predicted maps, we develop an Adversarial Integrity Learning (AIL) strategy to make the segmented regions more likely composed of complete objects with less surrounding noise. Extensive experiments on all the latest CoSOD benchmarks demonstrate that our lite MCCL outperforms 13 cutting-edge models, achieving the new state of the art (∼5.9% and ∼6.2% improvement in S-measure on CoSOD3k and CoSal2015, respectively). Our source codes, saliency maps, and online demos are publicly available at https://github.com/ZhengPeng7/MCCL.

GIONet: Global information optimized network for multi-center COVID-19 diagnosis via COVID-GAN and domain adversarial strategy

The outbreak of coronavirus disease (COVID-19) in 2019 has highlighted the need for automatic diagnosis of the disease, which can develop rapidly into a severe condition. Nevertheless, distinguishing between COVID-19 pneumonia and community-acquired pneumonia (CAP) through computed tomography scans can be challenging due to their similar characteristics. The existing methods often perform poorly in the 3-class classification task of healthy, CAP, and COVID-19 pneumonia, and they have poor ability to handle the heterogeneity of multi-centers data. To address these challenges, we design a COVID-19 classification model using global information optimized network (GIONet) and cross-centers domain adversarial learning strategy. Our approach includes proposing a 3D convolutional neural network with graph enhanced aggregation unit and multi-scale self-attention fusion unit to improve the global feature extraction capability. We also verified that domain adversarial training can effectively reduce feature distance between different centers to address the heterogeneity of multi-center data, and used specialized generative adversarial networks to balance data distribution and improve diagnostic performance. Our experiments demonstrate satisfying diagnosis results, with a mixed dataset accuracy of 99.17% and cross-centers task accuracies of 86.73% and 89.61%.

SGAT: Snapshot-guided adversarial training of neural networks

It is well known that ensembles of different models can achieve better performance. Unfortunately, training multiple deep neural networks is time-consuming and expensive. Also, they require more space for storage and more computation for inference, making them unsuitable for applications with limited resources, such as mobile and embedded devices. To address these two fundamental problems in ensemble learning, we propose a novel method called Snapshot-Guided Adversarial Training (SGAT) to automatically accumulate and distill knowledge from the training, where knowledge from the earlier training iterations of the network is distilled, and transferred into its late training iterations via an adversarial learning strategy. To accumulate knowledge from the training process, we employ a cyclic annealing schedule and take a model snapshot at the end of each training interval. Furthermore, we use a shared discriminator to encourage the distillation process. The main advantages of our method are: (1) We can directly train a network with its free snapshots for knowledge distillation, instead of heavily depending on pre-trained models; (2) the inference cost remains the same as using a single model after finishing knowledge transferring; (3) SGAT is a general method and can be applied to the existing network architectures. Our extensive experiments show that SGAT consistently outperforms the standing training method with a clear margin. For example, with the same training budget, it achieves 2.86% more accuracy on average than the baseline when training MobileNet-V2 from scratch on CIFAR-100. Meanwhile, SGAT also achieves better performance in most cases than the existing ensemble method and knowledge distillation. For example, to train MobileNet-V2 from scratch on ImageNet, it gets 0.7% more accuracy than snapshot ensembles, and 0.93% more accuracy than snapshot distillation. More importantly, our accumulated learning strategy makes SGAT achieve much better performance when we increase training time. For example, compared with the standard training method for MobileNet-V2, it gets 3.13% more accuracy on ImageNet.

An industry perspective for algorithm performance, testing, and validation

It is well known that ensembles of different models can achieve better performance. Unfortunately, training multiple deep neural networks is time-consuming and expensive. Also, they require more space for storage and more computation for inference, making them unsuitable for applications with limited resources, such as mobile and embedded devices. To address these two fundamental problems in ensemble learning, we propose a novel method called Snapshot-Guided Adversarial Training (SGAT) to automatically accumulate and distill knowledge from the training, where knowledge from the earlier training iterations of the network is distilled, and transferred into its late training iterations via an adversarial learning strategy. To accumulate knowledge from the training process, we employ a cyclic annealing schedule and take a model snapshot at the end of each training interval. Furthermore, we use a shared discriminator to encourage the distillation process. The main advantages of our method are: (1) We can directly train a network with its free snapshots for knowledge distillation, instead of heavily depending on pre-trained models; (2) the inference cost remains the same as using a single model after finishing knowledge transferring; (3) SGAT is a general method and can be applied to the existing network architectures. Our extensive experiments show that SGAT consistently outperforms the standing training method with a clear margin. For example, with the same training budget, it achieves 2.86% more accuracy on average than the baseline when training MobileNet-V2 from scratch on CIFAR-100. Meanwhile, SGAT also achieves better performance in most cases than the existing ensemble method and knowledge distillation. For example, to train MobileNet-V2 from scratch on ImageNet, it gets 0.7% more accuracy than snapshot ensembles, and 0.93% more accuracy than snapshot distillation. More importantly, our accumulated learning strategy makes SGAT achieve much better performance when we increase training time. For example, compared with the standard training method for MobileNet-V2, it gets 3.13% more accuracy on ImageNet.

DGANDDI: Double Generative Adversarial Networks for Drug-Drug Interaction Prediction.

Co-administration of multiple drugs may cause adverse drug interactions and side effects that damage the body. Therefore, accurate prediction of drug-drug interaction (DDI) events is of great importance. Recently, many computational methods have been proposed for predicting DDI associated events. However, most existing methods merely considered drug associated attribute information or topological information in DDI network, ignoring the complementary knowledge between them. Therefore, to effectively explore the complementarity of drug attribute and topological information of DDI network, we propose a deep learning model based adversarial learning strategy, which is named as DGANDDI. In DGANDDI, we design a two-GAN architecture to deeply capture the complementary knowledge between drug attribute and topological information of DDI network, thus more comprehensive drug representations can be learned. We conduct extensive experiments on real world dataset. The experimental results show that DGANDDI can effectively predict DDI occurrence and outperforms the comparison of the state-of-the-art models. We also perform ablation studies that demonstrate that DGANDDI is effective and that it is robust in DDI prediction tasks, even in the case of a scarcity of labeled DDIs.

Domain adaptation towards speaker-independent ultrasound tongue imaging based articulatory-to-acoustic conversion

In this paper, we endeavor to address an articulatory-to-acoustic issue which aims to estimate the mel-spectrogram of the acoustical signals, using midsagittal ultrasound tongue images of the vocal tract as input. Previous attempts employed statistical methods for the inversion between the articulatory movements and speech, while deep learning has begun to dominate this field. Despite the sustainable efforts that have been made, the mapping performance can be greatly varied for different speakers and most of the previous methods are constrained for the speaker-dependent scenario. Here, we present a novel approach towards speaker-independent mapping, which is inspired by the domain adaptation method. Specifically, we explore decoupling the spectrogram generation task and the speaker recognition task. Leveraging a novel designed loss function, we can improve the performance under the speaker-independent scenarios, through the adversarial learning strategy. To demonstrate the effectiveness of the proposed method, extensive experiments are conducted on the Tongue and Lips (TaL) corpus. Objective evaluation is conducted to compare the generated spectrograms and ground truth, using three evaluation metrics, including the MSE, SSIM, and CW-SSIM. The results indicate that our proposed method can achieve superior performance under the speaker-independent scenario, compared with competitive solutions. Our code is available at https://github.com/xianyi11/Articulatory-to-Acoustic-with-Domain-Adaptation.

Privacy-Preserving Cross-Environment Human Activity Recognition.

Recent studies have demonstrated the success of using the channel state information (CSI) from the WiFi signal to analyze human activities in a fixed and well-controlled environment. Those systems usually degrade when being deployed in new environments. A straightforward solution to solve this limitation is to collect and annotate data samples from different environments with advanced learning strategies. Although workable as reported, those methods are often privacy sensitive because the training algorithms need to access the data from different environments, which may be owned by different organizations. We present a practical method for the WiFi-based privacy-preserving cross-environment human activity recognition (HAR). It collects and shares information from different environments, while maintaining the privacy of individual person being involved. At the core of our approach is the utilization of the Johnson-Lindenstrauss transform, which is theoretically shown to be differentially private. Based on that, we further design an adversarial learning strategy to generate environment-invariant representations for HAR. →doWe demonstrate the effectiveness of the proposed method with different data modalities from two real-life environments. More specifically, on the raw CSI dataset, it shows 2.18% and 1.24% improvements over challenging baselines for two environments, respectively. Moreover, with the discrete wavelet transform features, it further yields 5.71% and 1.55% improvements, respectively.

Robust reconstruction of fluorescence molecular tomography based on adaptive adversarial learning strategy

Objective. Fluorescence molecular tomography (FMT) is a promising molecular imaging modality for quantifying the three-dimensional (3D) distribution of tumor probes in small animals. However, traditional deep learning reconstruction methods that aim to minimize the mean squared error (MSE) and iterative regularization algorithms that rely on optimal parameters are typically influenced by strong noise, resulting in poor FMT reconstruction robustness. Approach. In this letter, we propose an adaptive adversarial learning strategy (3D-UR-WGAN) to achieve robust FMT reconstructions. Unlike the pixel-based MSE criterion in traditional CNNs or the regularization strategy in iterative solving schemes, the reconstruction strategy can greatly facilitate the performance of the network models through alternating loop training of the generator and the discriminator. Second, the reconstruction strategy combines the adversarial loss in GANs with the L1 loss to significantly enhance the robustness and preserve image details and textual information. Main results. Both numerical simulations and physical phantom experiments demonstrate that the 3D-UR-WGAN method can adaptively eliminate the effects of different noise levels on the reconstruction results, resulting in robust reconstructed images with reduced artifacts and enhanced image contrast. Compared with the state-of-the-art methods, the proposed method achieves better reconstruction performance in terms of target shape recovery and localization accuracy. Significance. This adaptive adversarial learning reconstruction strategy can provide a possible paradigm for robust reconstruction in complex environments, and also has great potential to provide an alternative solution for solving the problem of poor robustness encountered in other optical imaging modalities such as diffuse optical tomography, bioluminescence imaging, and Cherenkov luminescence imaging.

Multi-sensor open-set cross-domain intelligent diagnostics for rotating machinery under variable operating conditions

Domain adaptation techniques have the proven ability to deal with fault diagnosis issues under variable operating conditions. They can achieve a superb diagnostic performance in single-sensor monitoring scenarios where the training and test data share the same label space. However, in practical engineering, fault modes are usually mixed with each other and new failure modes may appear during operation, which poses a challenge to the effectiveness of existing cross-domain fault diagnosis methods. Furthermore, with the increasing complexity of modern industrial systems, multi-sensor collaborative monitoring has been increasingly deployed for comprehensive measurement and detection of the complicated system. Unfortunately, there is less attention paid to multi-sensor cross-domain diagnosis in the current literature. To bridge this research gap, this paper aims to develop a novel multi-sensor open-set cross-domain fault diagnosis method. First, a convolutional neural network-based single-sensor feature extraction module and a Transformer-based multi-sensor feature fusion module are constructed for discriminative feature extraction and fusion. Second, a weighted adversarial learning scheme is built to conduct domain-invariant learning of the shared fault modes between the source and target domains. Then, a threshold-based supervised contrastive loss is defined to realize instance-level domain alignment, together with an entropy max–min loss to identify unknown class samples. The effectiveness and practicability of the proposed method are validated by a series of experiments conducted on two different types of gearboxes.

Stacking sequence optimization of arbitrary quadrilateral laminated plates for maximum fundamental frequency by hybrid whale optimization algorithm

A Hybrid Whale Optimization Algorithm (HWOA) is proposed to optimize the stacking sequence of arbitrary quadrilateral composite plates. By introducing an adversarial learning strategy, a broader search is performed on the initial domain to generate a better initial population. The mutation operator and opposition-based learning (OBL) strategy of the differential evolution (DE) algorithm are combined to maintain the diversity of the evolutionary process. In addition, the nonlinear convergence factor is introduced to improve the global exploration ability of HWOA when dealing with discrete problems. The stacking sequences of quadrilateral composite plates with different boundary conditions and geometries are optimized to maximize the fundamental frequency. The optimized results of HWOA are compared with those of whale optimization algorithm (WOA), Aquila optimizer (AO), artificial bee colony (ABC) algorithm, DE algorithm and layerwise optimization approach (LOA) algorithm. The results show that HWOA with multiple strategy combinations has good global search capability and robustness, and thus HWOA is a good candidate optimization algorithm for composite plate optimization.

Recognition-Oriented Image Compressive Sensing With Deep Learning

A number of image compressive sensing (CS) algorithms were proposed in the past two decades, aiming at yielding recovered images with the best possible visual effect. However, it is quite difficult to further improve the image quality for human eyes. For example, in the low-rate sampling scenarios, CS algorithms always suffer degraded performance and can only recover less visually appealing images. We notice that what human beings concern with is the visual quality of an image, while machine users care much more about its latent metrics, such as recognition accuracy, rather than the subjective visual effect. Inspired by this point, we develop a machine recognition-oriented image CS with an adversarial learning strategy. Some adversarial models are investigated to make the recognition accuracy as an additional optimization goal of the CS reconstruction network. Through end-to-end training, CS reconstruction network automatically learns an image recognition pattern, and produce recovered images owning extra recognition metric, which makes them become more suited for machine users. Experimental results indicate that the images recovered with the proposed adversarial learning strategy can be recognized with significantly higher accuracy compared to that with the existing CS algorithms.