Source Domain Data Research Articles

Human Activity Recognition based on Local Linear Embedding and Geodesic Flow Kernel on Grassmann manifolds

Human Activity Recognition (HAR) plays a crucial role in various applications(e.g., medical treatment, video surveillance and sports monitoring). Transfer learning is a promising solution to cross-domain identification problems in HAR. However, existing methods usually ignore the negative transfer caused by using the features of each source domain in equal proportions, as well as the distribution difference between the source and target domains. In this paper, an HAR method based on manifold learning is proposed. Firstly, the similarity between the domain and multiple source domains is calculated using the Multi-Kernel-Maximum Mean Difference (MK-MMD), and the source domain most similar to the target domain is selected as the optimal source domain in the transfer task. Secondly, Locally Linear Embedding (LLE) is leveraged to reduce the dimensionality of both optimal source domain and target domain data to remove redundant information, and the Geodesic Flow Kernel (GFK) is utilized to project low-dimensional data into the Grassmann manifold space and reduce the distribution difference between the two domains. Finally, the source domain action training model is applied to the target domain. Three public datasets (i.e., PAMAP2, OPPORTUNITY and UCI DSADS) are utilized to validate the effectiveness of the proposed approach. Experimental results are presented to demonstrate that the proposed HAR method can predict a large number of unlabeled samples in the target domain while preserving the original data structure.

Unsupervised multi-source domain adaptation for person re-identification via sample weighting

The aim of unsupervised domain adaptation (UDA) in person re-identification (re-ID) is to develop a model that can identify the same individual across different cameras in the target domain, using labeled data from the source domain and unlabeled data from the target domain. However, existing UDA person re-ID methods typically assume a single source domain and a single target domain, and seldom consider the scenario of multiple source domains and a single target domain. In the latter scenario, differences in sample size between domains can lead to biased training of the model. To address this, we propose an unsupervised multi-source domain adaptation person re-ID method via sample weighting. Our approach utilizes multiple source domains to leverage valuable label information and balances the inter-domain sample imbalance through sample weighting. We also employ an adversarial learning method to align the domains. The experimental results, conducted on four datasets, demonstrate the effectiveness of our proposed method.

MLN-net: A multi-source medical image segmentation method for clustered microcalcifications using multiple layer normalization

The accurate segmentation of clustered microcalcifications in mammography is crucial for the diagnosis and treatment of breast cancer. Despite exhibiting expert-level accuracy, recent deep learning advancements in medical image segmentation provide insufficient contribution to practical applications, due to the domain shift resulting from differences in patient postures, individual gland density, and imaging modalities, etc. In this paper, a novel framework named MLN-net is proposed for clustered microcalcification segmentation. It can segment multi-source images using only single source images. Specifically, to rich domain distribution information, we introduce a source domain image augmentation for generating multi-source images. A structure of multiple layer normalization (LN) layers is then used to construct the segmentation network, which can be found efficient for clustered microcalcification segmentation in different domains. Additionally, a branch selection strategy is designed for measuring the similarity of the source domain data and the target domain data. To validate the proposed MLN-net, extensive analyses including ablation experiments are performed, comparison of 12 baseline methods. MLN-net enhances segmentation quality of full-field digital mammography (FFDM) and digital breast tomosynthe (DBT) images from the FFDM-DBT dataset, achieving the average Dice similarity coefficient (DSC) of 86.52% and the average Hausdorff distance (HD) of 20.49 mm on the source domain DBT. And it outperforms the baseline models for the task in FFDM images from both the CBIS-DDSM and the FFDM-DBT dataset, achieving the average DSC of 50.78% and the average HD of 35.12 mm on the source domain CBIS-DDSM. Extensive experiments validate the effectiveness of MLN-net in segmenting clustered microcalcifications from different domains and its segmentation accuracy surpasses state-of-the-art methods. Code will be available at https://github.com/yezanting/MLN-NET-VERSON1.

Novel dual-network autoencoder based adversarial domain adaptation with Wasserstein divergence for fault diagnosis of unlabeled data

Intelligent fault diagnostic techniques based on deep learning have been developed by leaps and bounds, but it is quite difficult to construct a good fault diagnosis model without obtaining sufficient fault data labels. To solve this problem, this article proposes a novel dual-network autoencoder based adversarial domain adaptation with Wasserstein divergence (DWADA). Firstly, a dual-network autoencoder composed of the convolutional neural network (CNN) and the long short-term memory (LSTM) is regarded as a feature extractor for extracting local deep features and adding temporal feature information, while unsupervised reconstruction of the source domain data can ensure high accuracy of the classifier. Secondly, the domain discriminator forms an adversarial training with the feature extractor to facilitate the feature extractor to extract domain-invariant features for classification by minimizing the Wasserstein distance that measures the difference in feature distribution between different domains. Finally, Wasserstein divergence is introduced to the adversarial process to remove the k-Lipschitz constraint for improving the stability of the training. The Tennessee Eastman process (TEP) and the industrial three-phase flow process (TPFP) are applied to verify the performance of DWADA. Simulation results show that DWADA outperforms other related methods in transfer fault diagnosis tasks under different operating conditions.

IDSN: A one-stage interpretable and differentiable STFT domain adaptation network for traction motor of high-speed trains cross-machine diagnosis

A surge of transfer fault diagnosis techniques has been proposed to guarantee the safe operation of traction motor systems. However, existing efforts highly depend on the availability of fault data in source domain, which is rare in practice due to the regular maintenance. Fortunately, self-customized testbeds provide an opportunity to easily obtain fault data, assuming that the simulated data can be utilized to monitor the real-world traction motor systems via the cross-machine diagnosis method. Besides, current deep learning-based cross-machine fault diagnosis methods suffer from the poor physical interpretability and the troublesome hype-parameter selection. To tackle aforementioned issues, a one-stage Interpretable and Differentiable STFT cross-machine dual-driven adaptation Network (IDSN) is proposed. In IDSN, a new paradigm termed interpretable differentiable STFT layer is devised, where a derivable coefficient is introduced to adjust pivotal parameters of STFT such as window length by the gradient descent. Prominently, it is a plug-and-play module, which can be embedded into the arbitrary typical network without conflict. Besides, a novel adaptive trade-off coefficient is developed to tackle the weight matching of the domain discrepancy metric. Finally, to ensure the reliability and effectiveness of cross-machine diagnosis, a concise yet valid smoothed joint maximum mean discrepancy is proposed, which simultaneously promotes intra-class compactness and inter-class separability. The results of experiments confirm that the proposed IDSN outperforms the state of the art.

Feature mixing and self-training for source-free active domain adaptation

Recently, active domain adaptation (ADA) has become an emerging paradigm, as it significantly boosts the performance of the model with additional supervision. Despite the impressive performance of ADA methods, they are assumed that the source domain can be available during model adaptation. In many scenarios, the source domain can be unavailable due to issues such as data privacy. Consequently, ADA methods fail to achieve the expected results or even become invalid. To solve the problem, we design an effective framework named Feature Mixing and Self-Training (FMAS) for source-free active domain adaptation. Specifically, we adopt Clustering-based Alignment to make the classification boundary clearer to be beneficial to the subsequent sample selection, which can also achieve initial domain adaptation. When reaching active sample selection, we utilize feature mixing to explore the uncertainty and diversity of the sample. We construct the sample candidate pool by Mixup-based Uncertainty Selection and then consider the diversity to overcome the redundancy among the pool by Entropy-based Diversity Selection. Finally, for the remaining unlabeled samples, we adopt the self-training framework, which screens out high-confidence samples to explore the information of unlabeled samples. Thorough experiments show that FMAS improves average accuracy by at least 1% compared to other methods on benchmark datasets OfficeHome and MiniDomainNet. And FMAS can also achieve better results with ADA methods without the source domain data, which demonstrates FMAS can outperform previous ADA methods.

On the Importance of Attention and Augmentations for Hypothesis Transfer in Domain Adaptation and Generalization.

Unsupervised domain adaptation (UDA) aims to mitigate the performance drop due to the distribution shift between the training and testing datasets. UDA methods have achieved performance gains for models trained on a source domain with labeled data to a target domain with only unlabeled data. The standard feature extraction method in domain adaptation has been convolutional neural networks (CNNs). Recently, attention-based transformer models have emerged as effective alternatives for computer vision tasks. In this paper, we benchmark three attention-based architectures, specifically vision transformer (ViT), shifted window transformer (SWIN), and dual attention vision transformer (DAViT), against convolutional architectures ResNet, HRNet and attention-based ConvNext, to assess the performance of different backbones for domain generalization and adaptation. We incorporate these backbone architectures as feature extractors in the source hypothesis transfer (SHOT) framework for UDA. SHOT leverages the knowledge learned in the source domain to align the image features of unlabeled target data in the absence of source domain data, using self-supervised deep feature clustering and self-training. We analyze the generalization and adaptation performance of these models on standard UDA datasets and aerial UDA datasets. In addition, we modernize the training procedure commonly seen in UDA tasks by adding image augmentation techniques to help models generate richer features. Our results show that ConvNext and SWIN offer the best performance, indicating that the attention mechanism is very beneficial for domain generalization and adaptation with both transformer and convolutional architectures. Our ablation study shows that our modernized training recipe, within the SHOT framework, significantly boosts performance on aerial datasets.

Novel cross-domain fault diagnosis method based on model-agnostic meta-learning embedded in adaptive threshold network

Fault diagnosis of rotating machinery based on deep learning requires many identical labeled data to achieve satisfactory results. However, in engineering scenarios, few typical fault data have been manually marked, resulting in a poor on-site diagnosis consequence. Thus, training an efficient cross-domain diagnosis model under limited data is extremely challenging. This paper proposes a novel cross-domain fault diagnosis method based on model-agnostic meta-learning embedded in Adaptive threshold network (ATN-MAML). The threshold acquisition module is built to determine the signal's irrelevant information threshold adaptively. The designed threshold function is embedded in the network to filter out invalid information effectively as a non-linear layer. Then, MAML's training strategy is optimized to prevent overfitting. Finally, many cross-domain comparison experiments are conducted to verify the effectiveness of the proposed method on two datasets. The results show the model can maintain high accuracy even when the source and target domain data are from different devices.

Mitigation of network security attacks in wireless multimedia sensors networks using intrusion detection system

<p>In this work, research propose a deep transfer learning-based Lecun network P-Lenet technique to move the knowledge contained in source domain data to the target domain and integrate to produce an efficient intrusion detection system (IDS) that would enhance the detection accuracy for any wireless multimedia sensors networks (WMSN) ecosystems. To prevent attacks on the software defined network (SDN) platform in real time, the proposed method placed a strong emphasis on anomaly detection as the primary mechanism. The Lecun network (LeNet) was investigated in this work, and a new variant of the network called the Lenet was proposed. In addition, research make use of techniques like as feature normalization to increase the accuracy of the algorithm predictions and to optimize the training process in such a way that it consumes the least amount of time and resources. The performance of the proposed model that was recommended was superior to that of existing network intrusion detection system (NIDS) algorithms.</p>

Unsupervised Vehicle Re-Identification Method Based on Source-Free Knowledge Transfer

The unsupervised domain-adaptive vehicle re-identification approach aims to transfer knowledge from a labeled source domain to an unlabeled target domain; however, there are knowledge differences between the target domain and the source domain. To mitigate domain discrepancies, existing unsupervised domain-adaptive re-identification methods typically require access to source domain data to assist in retraining the target domain model. However, for security reasons, such as data privacy, data exchange between different domains is often infeasible in many scenarios. To this end, this paper proposes an unsupervised domain-adaptive vehicle re-identification method based on source-free knowledge transfer. First, by constructing a source-free domain knowledge migration module, the target domain is consistent with the source domain model output to train a generator to generate the “source-like samples”. Then, it can effectively reduce the model knowledge difference and improve the model’s generalization performance. In the experiment, two mainstream public datasets in this field, VeRi776 and VehicleID, are tested experimentally, and the obtained rank-k (the cumulative matching features) and mAP (the mean Average Precision) indicators are both improved, which are suitable for object re-identification tasks when data between domains cannot be interoperated.

Tensor representation-based transferability analytics and selective transfer learning of prognostic knowledge for remaining useful life prediction across machines

In recent years, deep transfer learning techniques have been successfully applied to solve RUL prediction across different working conditions. However, for RUL prediction across different machines in which the data distribution and fault evolution characteristics vary largely, the extraction and transition of prognostic knowledge become more challenging. Even if fault mode information can assist in the knowledge transfer, model bias will inevitably exist on the target machine with mixed or unknown faults. To address this issue from a transferability perspective, this paper proposes a novel selective transfer learning approach for RUL prediction across machines. First, the paper utilizes the tensor representation to construct the meta-degradation trend of each fault mode and evaluates the transferability of source domain data from fault mode and degradation characteristics through a new cross-machine transfer degree indicator (M-TDI). Second, a Long Short-Term Memory (LSTM)-based selective transfer strategy is proposed using the M-TDIs. The paper designs a training algorithm with an alternating optimization scheme to seek the optimal tensor decomposition and knowledge transfer effect. Theoretical analysis proves that the proposed approach significantly reduces the upper bound of prediction error. Furthermore, experimental results on three benchmark datasets prove the effectiveness of the proposed approach.

Source-Free Domain Adaptation via Target Prediction Distribution Searching

Existing Source-Free Domain Adaptation (SFDA) methods typically adopt the feature distribution alignment paradigm via mining auxiliary information (eg., pseudo-labelling, source domain data generation). However, they are largely limited due to that the auxiliary information is usually error-prone whilst lacking effective error-mitigation mechanisms. To overcome this fundamental limitation, in this paper we propose a novel Target Prediction Distribution Searching (TPDS) paradigm. Theoretically, we prove that in case of sufficient small distribution shift, the domain transfer error could be well bounded. To satisfy this condition, we introduce a flow of proxy distributions that facilitates the bridging of typically large distribution shift from the source domain to the target domain. This results in a progressive searching on the geodesic path where adjacent proxy distributions are regularized to have small shift so that the overall errors can be minimized. To account for the sequential correlation between proxy distributions, we develop a new pairwise alignment with category consistency algorithm for minimizing the adaptation errors. Specifically, a manifold geometry guided cross-distribution neighbour search is designed to detect the data pairs supporting the Wasserstein distance based shift measurement. Mutual information maximization is then adopted over these pairs for shift regularization. Extensive experiments on five challenging SFDA benchmarks show that our TPDS achieves new state-of-the-art performance. The code and datasets are available at https://github.com/tntek/TPDS.

Improving domain generalization performance for medical image segmentation via random feature augmentation.

Deep convolutional neural networks (DCNNs) have shown remarkable performance in medical image segmentation tasks. However, medical images frequently exhibit distribution discrepancies due to variations in scanner vendors, operators, and image quality, which pose significant challenges to the robustness of trained models when applied to unseen clinical data. To address this issue, domain generalization methods have been developed to enhance the generalization ability of DCNNs. Feature space-based data augmentation methods have been proven effective in improving domain generalization, but they often rely on prior knowledge or assumptions, which can limit the diversity of source domain data. In this study, we propose a novel random feature augmentation (RFA) method to diversify source domain data at the feature level without prior knowledge. Specifically, our RFA method perturbs domain-specific information while preserving domain-invariant information, thereby adequately diversifying the source domain data. Furthermore, we propose a dual-branches invariant synergistic learning strategy to capture domain-invariant information from the augmented features of RFA, enabling DCNNs to learn a more generalized representation. We evaluate our proposed method on two challenging medical image segmentation tasks, optic cup/disc segmentation on fundus images and prostate segmentation on MRI images. Extensive experimental results demonstrate the superior performance of our method over state-of-the-art domain generalization methods.

Cluster-based Dual-branch Contrastive Learning for unsupervised domain adaptation person re-identification

Unsupervised domain adaptation (UDA) person re-identification (Re-ID) is to enhance the discriminability of Re-ID tasks in the target domain by leveraging labeled source domain data and unlabeled target domain data. Employing clustering as a mean to generate pseudo labels has emerged as a successful approach for addressing UDA Re-ID tasks. However, clustering may be disturbed by data noise when generating pseudo labels, resulting in unstable quality of pseudo labels. In addition, persons with similar colored clothing can also affect the recognition accuracy of the model. To address these aforementioned issues in UDA Re-ID, we propose a Cluster-based Dual-branch Contrastive Learning (CDCL) approach. In CDCL, we design partially grayed images as input and utilize two-branch network structure to reduce the effect of upper-body clothing color on model recognition. Based on the two-branch network architecture, we additionally propose a cluster refinement method to improve the confidence of pseudo labels, which selects samples with high similarity, i.e., samples with weighted distance between subclass samples less than a threshold, and thus makes pseudo labels more reliable. The proposed CDCL has shown its effectiveness through experimental results, achieving an mAP accuracy of 81.5%.

M-MSSEU: source-free domain adaptation for multi-modal stroke lesion segmentation using shadowed sets and evidential uncertainty.

Due to the unavailability of source domain data encountered in unsupervised domain adaptation, there has been an increasing number of studies on source-free domain adaptation (SFDA) in recent years. To better solve the SFDA problem and effectively leverage the multi-modal information in medical images, this paper presents a novel SFDA method for multi-modal stroke lesion segmentation in which evidential deep learning instead of convolutional neural network. Specifically, for multi-modal stroke images, we design a multi-modal opinion fusion module which uses Dempster-Shafer evidence theory for decision fusion of different modalities. Besides, for the SFDA problem, we use the pseudo label learning method, which obtains pseudo labels from the pre-trained source model to perform the adaptation process. To solve the unreliability of pseudo label caused by domain shift, we propose a pseudo label filtering scheme using shadowed sets theory and a pseudo label refining scheme using evidential uncertainty. These two schemes can automatically extract unreliable parts in pseudo labels and jointly improve the quality of pseudo labels with low computational costs. Experiments on two multi-modal stroke lesion datasets demonstrate the superiority of our method over other state-of-the-art SFDA methods.

Smart filter aided domain adversarial neural network for fault diagnosis in noisy industrial scenarios

The application of unsupervised domain adaptation (UDA)-based fault diagnosis methods has shown significant efficacy in industrial settings, facilitating the transfer of operational experience and fault signatures between different operating conditions, different units of a fleet or between simulated and real data. However, in real industrial scenarios, unknown levels and types of noise can amplify the difficulty of domain alignment, thus severely affecting the diagnostic performance of deep learning models. To address this issue, we propose an UDA method called Smart Filter-Aided Domain Adversarial Neural Network (SFDANN) for fault diagnosis in noisy industrial scenarios. The proposed methodology comprises two steps. In the first step, we develop a smart filter that dynamically enforces similarity between the source and target domain data in the time-frequency domain. This is achieved by combining a learnable wavelet packet transform network (LWPT) and a traditional wavelet packet transform module. In the second step, we input the data reconstructed by the smart filter into a domain adversarial neural network (DANN). To learn domain-invariant and discriminative features, the learnable modules of SFDANN are trained in a unified manner with three objectives: time-frequency feature proximity, domain alignment, and fault classification. We validate the effectiveness of the proposed SFDANN method based on two fault diagnosis cases: one involving fault diagnosis of bearings in noisy environments and another involving fault diagnosis of slab tracks in a train-track-bridge coupling vibration system, where the transfer task involves transferring from numerical simulations to field measurements. Results show that compared to other representative state of the art UDA methods, SFDANN exhibits superior performance and remarkable stability.

An soft-sensor method for the biochemical reaction process based on LSTM and transfer learning

Due to significant differences in data distribution under different working conditions during Pichia pastoris biochemical reaction process, traditional soft-sensor model suffer from the model failure and deterioration, this paper propose a soft-sensor modeling method combing long short-term memory network (LSTM) and balanced distribution adaptation method (BDA). Firstly, the source domain data is used to establish an accurate source domain LSTM prediction model, and the structure and parameters of the first layer of LSTM are fixed to migrate to the target domain prediction model. Then use the balanced distribution adaptation method to shrink the distribution differences between different domains of data. Finally, data that has been modeled with balanced and adaptive distribution assist the real-time data to train the remaining layer of the network, and the accurate target domain prediction model is finally obtained. The simulation results show that the mentioned method has the preponderance of timely prediction and high prediction accuracy, which validates the effectiveness and practicality of the method. This method solves the problem of soft-sensor modeling under unknown modes of multiple operating conditions in Pichia pastoris biochemical reaction process, achieving the prediction of key parameters under different operating conditions, which can be widely applied, and also providing a new method for soft-sensor modeling of other non system systems.

Raman spectrum model transfer method based on Cycle-GAN

The disparity in hardware quality among various models of Raman spectrometers gives rise to variations in the acquired Raman spectral data, even when the same substance is collected under identical external conditions. Conventionally, models constructed using data obtained from a particular instrument exhibit issues such as limited applicability or poor performance when deployed to different instruments. Currently, numerous model transfer algorithms grounded in chemometrics have been developed, all aiming to establish a mapping relationship capable of transforming spectral data from the source domain to the target domain. With the advancement of deep learning techniques, the utilization of deep learning enables the effective resolution of nonlinear mapping relationships between two spectral vectors. In the field of image translation, the Cycle-Consistent Adversarial Networks, Cycle-GAN, has already achieved mutual transformation between two distinct style images. However, due to images being multidimensional matrix data, unlike one-dimensional spectral data vectors, we have constructed a deep learning network based on Cycle-GAN for vector-to-vector transformation. This network allows the direct conversion of spectral data from the source domain to the target domain, without requiring parameter adjustments or other operations. Compared with traditional chemometric methods, our method is more intelligent and efficient. Finally, the cosine similarity between the source domain data and the transformed target domain data exceeds 99%.

Single-source UDA for privacy-preserving intelligent fault diagnosis based on domain augmentation

In practical applications of fault diagnosis, several factors, including fluctuations in load, changes in equipment condition, and environmental noise effects, could cause a classifier that’s been trained on the source domain to be ill-suited for matching data from the target domain. Unsupervised domain adaptation techniques have been developed to tackle this issue, but they typically demand access to fully labeled source domains, ignoring concerns of privacy regarding source domain data. Therefore, we consider a new research scene for source-free unsupervised domain adaptation (SFUDA), which exclusively relies on a source model trained on source domain sample without requiring access to fully labeled source domain data. This paper introduces a SFUDA approach that utilizes knowledge distillation (KD), which involves two stages: (1) generalizing the source model by applying domain augmentation techniques and LS methods that enhance the model’s potential to enhance its generalization capability; (2) adapting the target model using a KD framework to achieve knowledge migration; and in addition, mutual information structure regularization is added to consider the internal data structure, thus enhancing the model’s adaptability. To evaluate the efficacy of our approach, we perform experiments on two datasets—the Case Western Reserve University dataset and the Paderborn University dataset, comprising 24 transfer tasks. Our experiments demonstrate the effectiveness of the domain augmentation technique, mutual information regularization, and the proposed method.

Hyperspectral Classification of Frost Damage Stress in Tomato Plants Based on Few-Shot Learning

Early detection and diagnosis of crop anomalies is crucial for enhancing crop yield and quality. Recently, the combination of machine learning and deep learning with hyperspectral images has significantly improved the efficiency of crop detection. However, acquiring a large amount of properly annotated hyperspectral data on stressed crops requires extensive biochemical experiments and specialized knowledge. This limitation poses a challenge to the construction of large-scale datasets for crop stress analysis. Meta-learning is a learning approach that is capable of learning to learn and can achieve high detection accuracy with limited training samples. In this paper, we introduce meta-learning to hyperspectral imaging and crop detection for the first time. In addition, we gathered 88 hyperspectral images of drought-stressed tomato plants and 68 images of freeze-stressed tomato plants. The data related to drought serve as the source domain, while the data related to frost damage serve as the target domain. Due to the difficulty of obtaining target domain data from real-world testing scenarios, only a limited amount of target domain data and source domain data are used for model training. The results indicated that meta-learning, with a minimum of eight target domain samples, achieved a detection accuracy of 69.57%, precision of 59.29%, recall of 66.32% and F1-score of 62.61% for classifying the severity of frost stress, surpassing other methods with a target domain sample size of 20. Moreover, for determining whether the plants were under stress, meta-learning, with a minimum of four target domain samples, achieved a detection accuracy of 89.1%, precision of 89.72%, recall of 93.08% and F1-score of 91.37% outperforming other methods at a target domain sample size of 20. The results show that meta-learning methods require significantly less data across different domains compared to other methods. The performance of meta-learning techniques thoroughly demonstrates the feasibility of rapidly detecting crop stress without the need for collecting a large amount of target stress data. This research alleviates the data annotation pressure for researchers and provides a foundation for detection personnel to anticipate and prevent potential large-scale stress damage to crops.