Domain Shift Research Articles

A Patch Diversity Transformer for Domain Generalized Semantic Segmentation.

Domain generalization (DG) is one of the critical issues for deep learning in unknown domains. How to effectively represent domain-invariant context (DIC) is a difficult problem that DG needs to solve. Transformers have shown the potential to learn generalized features, since the powerful ability to learn global context. In this article, a novel method named patch diversity Transformer (PDTrans) is proposed to improve the DG for scene segmentation by learning global multidomain semantic relations. Specifically, patch photometric perturbation (PPP) is proposed to improve the representation of multidomain in the global context information, which helps the Transformer learn the relationship between multiple domains. Besides, patch statistics perturbation (PSP) is proposed to model the feature statistics of patches under different domain shifts, which enables the model to encode domain-invariant semantic features and improve generalization. PPP and PSP can help to diversify the source domain at the patch level and feature level. PDTrans learns context across diverse patches and takes advantage of self-attention to improve DG. Extensive experiments demonstrate the tremendous performance advantages of the PDTrans over state-of-the-art DG methods.

Reducing semantic ambiguity in domain adaptive semantic segmentation via probabilistic prototypical pixel contrast

Domain adaptation aims to reduce the model degradation on the target domain caused by the domain shift between the source and target domains. Although encouraging performance has been achieved by combining contrastive learning with the self-training paradigm, they suffer from ambiguous scenarios caused by scale, illumination, or overlapping when deploying deterministic embedding. To address these issues, we propose probabilistic prototypical pixel contrast (PPPC), a universal adaptation framework that models each pixel embedding as a probability via multivariate Gaussian distribution to fully exploit the uncertainty within them, eventually improving the representation quality of the model. In addition, we derive prototypes from probability estimation posterior probability estimation which helps to push the decision boundary away from the ambiguity points. Moreover, we employ an efficient method to compute similarity between distributions, eliminating the need for sampling and reparameterization, thereby significantly reducing computational overhead. Further, we dynamically select the ambiguous crops at the image level to enlarge the number of boundary points involved in contrastive learning, which benefits the establishment of precise distributions for each category. Extensive experimentation demonstrates that PPPC not only helps to address ambiguity at the pixel level, yielding discriminative representations but also achieves significant improvements in both synthetic-to-real and day-to-night adaptation tasks. It surpasses the previous state-of-the-art (SOTA) by +5.2% mIoU in the most challenging daytime-to-nighttime adaptation scenario, exhibiting stronger generalization on other unseen datasets. The code and models are available at https://github.com/DarlingInTheSV/Probabilistic-Prototypical-Pixel-Contrast.

Multivariate prototype representation for domain-generalized incremental learning

Deep learning models often suffer from catastrophic forgetting when fine-tuned with samples of new classes. This issue becomes even more challenging when there is a domain shift between training and testing data. In this paper, we address the critical yet less explored Domain-Generalized Class-Incremental Learning (DGCIL) task. We propose a DGCIL approach designed to memorize old classes, adapt to new classes, and reliably classify objects from unseen domains. Specifically, our loss formulation maintains classification boundaries while suppressing domain-specific information for each class. Without storing old exemplars, we employ knowledge distillation and estimate the drift of old class prototypes as incremental training progresses. Our prototype representations are based on multivariate Normal distributions, with means and covariances continually adapted to reflect evolving model features, providing effective representations for old classes. We then sample pseudo-features for these old classes from the adapted Normal distributions using Cholesky decomposition. Unlike previous pseudo-feature sampling strategies that rely solely on average mean prototypes, our method captures richer semantic variations. Experiments on several benchmarks demonstrate the superior performance of our method compared to the state of the art.

A Foundation Language-Image Model of the Retina (FLAIR): encoding expert knowledge in text supervision

Foundation vision-language models are currently transforming computer vision, and are on the rise in medical imaging fueled by their very promising generalization capabilities. However, the initial attempts to transfer this new paradigm to medical imaging have shown less impressive performances than those observed in other domains, due to the significant domain shift and the complex, expert domain knowledge inherent to medical-imaging tasks. Motivated by the need for domain-expert foundation models, we present FLAIR, a pre-trained vision-language model for universal retinal fundus image understanding. To this end, we compiled 38 open-access, mostly categorical fundus imaging datasets from various sources, with up to 101 different target conditions and 288,307 images. We integrate the expert’s domain knowledge in the form of descriptive textual prompts, during both pre-training and zero-shot inference, enhancing the less-informative categorical supervision of the data. Such a textual expert’s knowledge, which we compiled from the relevant clinical literature and community standards, describes the fine-grained features of the pathologies as well as the hierarchies and dependencies between them. We report comprehensive evaluations, which illustrate the benefit of integrating expert knowledge and the strong generalization capabilities of FLAIR under difficult scenarios with domain shifts or unseen categories. When adapted with a lightweight linear probe, FLAIR outperforms fully-trained, dataset-focused models, more so in the few-shot regimes. Interestingly, FLAIR outperforms by a wide margin larger-scale generalist image-language models and retina domain-specific self-supervised networks, which emphasizes the potential of embedding experts’ domain knowledge and the limitations of generalist models in medical imaging. The pre-trained model is available at: https://github.com/jusiro/FLAIR.

Adaptive wavelet-VNet for single-sample test time adaptation in medical image segmentation.

In medical image segmentation, a domain gap often exists between training and testing datasets due to different scanners or imaging protocols, which leads to performance degradation in deep learning-based segmentation models. Given the high cost of manual labeling and the need for privacy protection, it is often challenging to annotate the testing (target) domain data for model fine-tuning or to collect data from different domains to train domain generalization models. Therefore, using only unlabeled target domain data for test-time adaptation (TTA) presents a more practical but challenging solution. To improve the segmentation accuracy of deep learning-based models on unseen datasets, and especially to enhance the efficiency and stability of TTA for individual samples from heterogeneous domains. In this study, we proposed to dynamically adapt a wavelet-VNet (WaVNet) to unseen target domains with a hybrid objective function, based on each unlabeled test sample during the test time. We embedded multiscale wavelet coefficients into a V-Net encoder and adaptively adjusted the spatial and spectral features according to the input, and the model parameters were optimized by three loss functions. We integrated a shape-aware loss to focus on the foreground segmentations, a Refine loss to correct the incomplete and noisy segmentations caused by domain shifts, and an entropy loss to promote the global consistency of the segmentations. We evaluated the proposed method on multidomain liver and prostate segmentation datasets to assess its advantages over other TTA methods. For the source domain model training of the liver dataset, we used 15 3D MR image samples for training and 5 for validation. Correspondingly, for the prostate dataset, we used 22 3D MR image samples for training and 7 for validation. In the target domain, we used a single 3D MR image sample for adaptation and testing. The total number of testing samples is 60 in the liver dataset (for 3 different domains) and 116 in the prostate dataset (for 6 different domains). The proposed method showed the highest segmentation accuracy among all methods, achieving a mean(±SD) Dice coefficient (DSC) of 78.10±5.23% and a mean 95th Hausdorff distance (HD95) of 15.52±5.84mm on the liver dataset; and a mean DSC of 80.02±3.89% and a mean HD95 of 9.18±3.47mm on the prostate dataset. The DSC is 11.67% (in absolute terms) and 15.27% higher than that of the baseline (no adaptation) method, for the liver and the prostate datasets, respectively. The proposed adaptive WaVNet enhanced the image segmentation accuracy from unseen domains during the test time via unsupervised learning and multi-objective optimization. It can benefit clinical applications where data are scarce or with changing data distributions, including online adaptive radiotherapy. The code will be released at: https://github.com/sanny1226/WaVNet.

Multi-scale feature reconstruction network for industrial anomaly detection

Unsupervised anomaly detection techniques, which operate without prior knowledge of anomalies, have garnered significant attention in industrial inspection due to their adaptability and generalization. Therefore, knowledge-based computer vision techniques have been broadly applied to identify unusual image patterns. However, real-time industrial applications present challenges such as limited anomalous samples, inadequate defect knowledge, and complex background textures. These factors lead to difficulties in accurately identifying defect regions, and conventional auto-encoder networks often struggle to overcome these issues.To address these limitations, we propose a multi-scale feature reconstruction (MSFR) network specifically designed for domain shift scenarios. Our approach employs a pyramidal vision transformer network (PVTN) to reconstruct multi-scale feature maps, capturing discriminative features at various scales. Additionally, a pre-trained module extracts multi-level features at the same scale, and a dedicated feature matching module enhances accuracy by improving the alignment probability between features. The MSFR strategy surpasses conventional auto-encoders by filtering pixel-level information at multiple depths.Empirical evaluations were conducted using benchmark datasets such as MVTec AD and AeBAD-S. Furthermore, an extensive ablation study demonstrates the effectiveness and viability of the proposed MSFR approach for industrial anomaly detection tasks. The experimental results show that the proposed model significantly outperforms recent approaches, making it highly suitable for real-world industrial applications, particularly in manufacturing.

Multisource partial domain adaptation for bearing fault diagnosis

Abstract Domain adaptation has been widely used in fault diagnosis and dealt with data distribution discrepancies, but the labels in source and target domains are usually assumed to be identical. The complexity of the working conditions, in reality, leads to the fact that the labels in target domains are often a subset of the labels in source domains. This special case is called the partial domain problem. However, most of the existing proposed methods for solving partial domain problems are limited to single-source-domain scenarios and fail to effectively integrate multisource knowledge. Hence, this study proposes a new approach of multisource domain obfuscation-subdomain alignment (MSDO-SA) for partial domain adaptation fault diagnosis in multisource domains. Through domain obfuscation, the multisource domains are converted into a single source domain. The subdomain alignment aims at improving the generalization ability and relevance of the model, and effectively alleviates the domain shift problem. Finally, multiple partial domain fault diagnosis tasks using the CWRU dataset validate the effectiveness, robustness, and superiority of the proposed method.

Prior knowledge-guided vision-transformer-based unsupervised domain adaptation for intubation prediction in lung disease at one week

Data-driven approaches have achieved great success in various medical image analysis tasks. However, fully-supervised data-driven approaches require unprecedentedly large amounts of labeled data and often suffer from poor generalization to unseen new data due to domain shifts. Various unsupervised domain adaptation (UDA) methods have been actively explored to solve these problems. Anatomical and spatial priors in medical imaging are common and have been incorporated into data-driven approaches to ease the need for labeled data as well as to achieve better generalization and interpretation. Inspired by the effectiveness of recent transformer-based methods in medical image analysis, the adaptability of transformer-based models has been investigated. How to incorporate prior knowledge for transformer-based UDA models remains under-explored. In this paper, we introduce a prior knowledge-guided and transformer-based unsupervised domain adaptation (PUDA) pipeline. It regularizes the vision transformer attention heads using anatomical and spatial prior information that is shared by both the source and target domain, which provides additional insight into the similarity between the underlying data distribution across domains. Besides the global alignment of class tokens, it assigns local weights to guide the token distribution alignment via adversarial training. We evaluate our proposed method on a clinical outcome prediction task, where Computed Tomography (CT) and Chest X-ray (CXR) data are collected and used to predict the intubation status of patients in a week. Abnormal lesions are regarded as anatomical and spatial prior information for this task and are annotated in the source domain scans. Extensive experiments show the effectiveness of the proposed PUDA method.

Self-adaptive fault diagnosis for unseen working conditions based on digital twins and domain generalization

In recent years, intelligent fault diagnosis based on domain adaptation has been used to address domain shifts in cyber–physical systems; however, the need for acquiring target data sufficiently limits their applicability to unseen working conditions. To overcome such limitations, domain generalization techniques have been introduced to enhance the capacity of fault diagnostic models to operate under unseen working conditions. Nevertheless, existing approaches assume access to extensive labeled training data from various source domains, posing challenges in real-world engineering scenarios due to resource constraints. Moreover, the absence of a mechanism for updating diagnostic models over time calls for the exploration of self-adaptive generalized diagnosis models that are capable of autonomous reconfiguration in response to new unseen working conditions. In such a context, this paper proposes a self-adaptive fault diagnosis system that combines several paradigms, namely Monitor-Analyze-Plan-Execute over a shared Knowledge (MAPE-K), Domain Generalization Network Models (DGNMs), and Digital Twins (DT). The MAPE-K loop enables run-time adaptation to dynamic industrial environments without human intervention. To address the scarcity of labeled training data, digital twins are used to generate supplementary data and continuously tune parameters to reflect the dynamics of new unseen working conditions. DGNM incorporates adversarial learning and a domain-based discrepancy metric to enhance feature diversity and generalization. The introduction of multi-domain data augmentation enhances feature diversity and facilitates learning correlations among multiple domains, ultimately improving the generalization of feature representations. The proposed fault diagnosis system has been evaluated on three publicly available rotating machinery datasets to demonstrate its higher performance in cross-work operation and cross-machine tasks compared to other state-of-the-art methods.

Envelope spectrum neural network with adaptive domain weight harmonization for intelligent bearing fault diagnosis under cross-machine scenarios

Accurate bearing fault diagnosis technology is highly important for ensuring the safe operation of mechanical equipment. Fault diagnosis methods can be roughly divided into signal processing-based methods (SPM) and data-driven methods (DDMs), which rely on physical knowledge and data knowledge, respectively. However, SPM cannot adapt to large data samples and dynamic parameter variations. DDMs did not consider physical representation consistency. To address the above issues, an envelope spectrum neural network (ESNN) is proposed for cross-machine bearing fault diagnosis. First, a deep transfer convolution network (DTCN) is constructed as the basic diagnostic framework. Second, a knowledge-to-model conversion strategy (KMC) is built to create ESNN, which utilizes equivalent convolution functions to construct the first two layers of DTCN, guiding the model to learn physical knowledge. Subsequently, an adaptive domain weight harmonization (ADWH) mechanism is proposed that can dynamically combine marginal distributions and joint distributions and automatically learn hidden features contained in physical and data knowledge, thus alleviating domain shift issues. Experimental evaluations are conducted using fault datasets from three different machines. The results show that, compared to models without knowledge guidance, ESNN achieves a 6.4% improvement in diagnostic accuracy. Compared to many advanced models, ESNN can achieve a maximum diagnostic accuracy improvement of 35.94%. The code of the ESNN can be found at https://github.com/John-520/ESNN.

Towards generalized face forgery detection with domain-robust representation learning

Face forgery detection is crucial for the security of digital identities. However, existing methods often struggle to generalize effectively to unseen domains due to the domain shift between training and testing data. We propose a Domain-robust Representation Learning (DRRL) method for generalized face forgery detection. Specifically, we observe that domain shifts in face forgery detection tasks are often caused by forgery differences and content differences between domain data, while the limitations of training data lead the model to overfit to these feature expressions in the seen domain. Therefore, DRRL enhances the model's generalization to unseen domains by first adding representative data representations to mitigate overfitting to seen data and then removing the features of expressed domain information to learn a robust, discriminative representation of domain variation. Data augmentation is achieved by stylizing sample representations and exploring representative new styles to generate rich data variants, with the Content-style Augmentation (CSA) module and Forgery-style Augmentation (FSA) module implemented for content and forgery expression, respectively. Based on this, the Content Decorrelation (CTD) module and Sensitive Channels Drop (SCD) module are used to remove content features irrelevant to forgery and domain-sensitive forgery features, encouraging the model to focus on clean and robust forgery features, thereby achieving the goal of learning domain-robust representations. Extensive experiments on five large-scale datasets demonstrate that our method exhibits advanced and stable generalization performance in practical scenarios.

Variational inference based adversarial domain adaptation

AbstractDeep learning methods demand enormous amounts of labeled data. Although collecting labeled data is a challenge but it can be accomplished with difficulty. Nevertheless, domain shift or bias may occur due to some conditions. Recollecting data under similar conditions is too expensive or impossible. Domain adaptation is an effective technique to deal with this problem. Among domain adaptation methods, adversarial learning models handled domain shifts well. Adversarial learning’s biggest issue is matching the features to the appropriate classes. In other words, domain adaptation may lead to negative transfer. Utilizing adversarial learning and Variational Auto-Encoder (VAE) cluster-forming capabilities, we propose a method that overcomes these limitations. Our structure attempts to generate a smooth latent representation based on both the target and the source data using a variational auto-encoder. Then the affected data are fed into an adversarial learning component. The component includes an encoder and a discriminator. The discriminator aims to identify the source and target encoders’ features. The target encoder uses inverted domain labels to mislead the discriminator. To emphasize the desired effect of VAE on domain adaptation, we test the model performance without adversarial training. In comparison to the most prevalent adversarial domain adaptation benchmarks, our method yields approving and comparable results.

MSS-JDA: Multi-source self-selected joint domain adaptation method based on cross-subject EEG emotion recognition

The electroencephalogram (EEG) is widely used to identify different emotional states. However, cross-subject emotion recognition based on EEG data is challenging due to physiological differences among participants. Many studies have proposed using multi-source domain adaptation methods to address cross-subject issues and have achieved promising results. multi-source domain adaptation involves constructing independent branches for EEG data from multiple subjects, using one-to-one domain adaptation, and extracting domain-specific features, followed by reasoning through multiple branches. However, a current practical obstacle is the increasing computational time cost as the number of source domains grows. Moreover, too many source domains may lead to domain shift issues. To address these challenges, we propose a novel emotion recognition method based on Multi-Source Domain Branch Self-Selected Joint Domain Adaptation (MSS-JDA). Initially, we construct a shared feature extractor for multiple source domain branches to extract common low-level features. Next, we use a one-to-one associated joint adaptation method to extract features specific to each domain. During training, our method of multi-source domain branch self-selection, guided by prior knowledge in the early training phase, prunes source domain branches that deviate significantly from the target domain, retaining only a certain number of source domain branches close to the target domain for continued training. Finally, domain-specific features are input into their respective classifiers to obtain N predictions, and their average is taken as the final result. Extensive experiments on the SEED and SEED-IV datasets demonstrate that the MSS-JDA model achieves accuracy and standard deviation scores of 93.78%, 3.39 and 78.93%, 7.39 respectively, surpassing most existing models. The experiments indicate that the proposed MSS-JDA method improves the accuracy of EEG emotion recognition classification, thus benefiting practical applications in EEG classification.

A novel deep domain adaptation method based on three-way decisions for identifying autistic patients

Combining machine learning with functional magnetic resonance imaging (fMRI) technology to build an effective autism recognition model has become a research hotspot. However, the performance of existing machine learning models is highly dependent on large-scale labelled data, and the cost of fMRI data acquisition and labelling limits the application of models in autism recognition. Although the availability of multisite data increases the sample size, the resolution, image contrast, and noise level of fMRI data acquired at different sites vary considerably due to the different scanning devices or scanning protocols used. The heterogeneity of fMRI data between sites leads to limited generalisation of the trained models. To tackle this problem, we propose a deep domain adaptation approach based on three-way decisions. First, we introduce a pseudolabel optimisation method based on three-way decisions, which fully considers the structural information of the target domain and makes secondary decisions on the uncertain objects in the target domain, thus improving the accuracy of the pseudolabel in the target domain. Then, we construct a new autism recognition network using a stacked autoencoder and combine the proposed pseudolabel optimisation model to adapt the domain shift. Finally, the self-training strategy is used to retrain the network after domain adaptation to improve the generalisability of the model to the target domain. Experimental results on several cross-domain autism recognition tasks show the effectiveness of the proposed method.

Learning Domain Invariant Features for Unsupervised Indoor Depth Estimation Adaptation

Predicting depth maps from monocular images has made an impressive performance in the past years. However, most depth estimation methods are trained with paired image-depth map data or multi-view images (e.g., stereo pair and monocular sequence), which suffer from expensive annotation costs and poor transferability. Although unsupervised domain adaptation methods are introduced to mitigate the reliance on annotated data, rare works focus on the unsupervised cross-scenario indoor monocular depth estimation. In this article, we propose to study the generalization of depth estimation models across different indoor scenarios in an adversarial-based domain adaptation paradigm. Concretely, a domain discriminator is designed for discriminating the representation from source and target domains, while the feature extractor aims to confuse the domain discriminator by capturing domain-invariant features. Further, we reconstruct depth maps from latent representations with the supervision of labeled source data. As a result, the feature extractor learned features possess the merit of both domain-invariant and low source risk, and the depth estimator can deal with the domain shift between source and target domains. We conduct the cross-scenario and cross-dataset experiments on the ScanNet and NYU-Depth-v2 datasets to verify the effectiveness of our method and achieve impressive performance.

DCST: Dual Cross-Supervision for Transformer-based Unsupervised Domain Adaptation

Unsupervised Domain Adaptation aims to leverage a source domain with ample labeled data to tackle tasks on an unlabeled target domain. However, this poses a significant challenge, particularly in scenarios exhibiting significant disparities between the two domains. Prior methods often fall short in challenging domains due to the impact of incorrect pseudo-labeling noise and the limits of handcrafted domain alignment rules. In this paper, we propose a novel method called DCST (Dual Cross-Supervision Transformer), which improves upon existing methods in two key aspects. Firstly, vision transformer is combined with dual cross-supervision learning strategy to enforce consistency learning from different domains. The network accomplishes domain-specific self-training and cross-domain feature alignment in an adaptive manner. Secondly, due to the presence of noise in challenging domain, and the need to reduce the risks of model collapse and overfitting, we propose a Domain Shift Filter. Specifically, this module allows the model to leverage the memory of source domain features to facilitate a smooth transition. It can also improve the effectiveness of knowledge transfer between domains with significant gaps. We conduct extensive experiments on four benchmark datasets and achieved the best classification results, including 94.3% on Office-31, 86.0% on Office-Home, 89.3% on VisDA-2017, and 48.8% on DomainNet. Code is available in https://github.com/Yislight/DCST.

Heterogeneous Transfer Learning of Electrohydrodynamic Printing Under Zero-Gravity Toward In-Space Manufacturing

Abstract As we continue to commercialize space and mature in-space manufacturing (ISM) processes, there is a strong need to transfer the knowledge we learn from experiments on the ground to zero-gravity environments. Physics-motivated manufacturing processes, like additive manufacturing, experience a shift in fabrication parameters due to the absence of gravity and the change of environments. Thus, we found traditional machine learning methods are not capable of addressing this domain shift and present a transfer learning scheme as a solution in this paper. We tested a kernel ridge regression model built for heterogeneous transfer learning (KRR-HeITL) on data from the electrohydrodynamic inkjet printing (EHD printing) process. EHD printing is a process that uses electrical force to control material flows, thus achieving the fabrication of electronics without requiring gravity. Our team has successfully conducted three rounds of parabolic flights to validate this technology for ISM. We trained on multiple datasets built from on-ground experiments and tested using zero-gravity printing data obtained from parabolic flight tests. Measurements of the Taylor cone both on-ground and in zero-gravity were taken and exploited as a part of the training data. We found that our method obtains good interpolation accuracy (MAPE 3.85%) compared to traditional machine learning methods (MAPE 16.84%) for predicting the printed line width. We concluded that the KRR-HeITL method is well suited for zero-gravity domain shifts of EHD printing parameters. This study paves the way for future predictions of ISM parameters when there are only on-ground experiments or very limited zero-gravity datasets for a given process.

Machine cross-domain remaining useful life prediction via contrastive adversarial variational recurrent method

The performance degradation process of the machine is non-stationary which varies with the operating environment and is reflected as a temporal shift phenomenon. Models or methods that assume the test and training sets have the same distribution will no longer be suitable for solving problems where domain shifts exist. This brings new challenges for accurately evaluating the machine’s remaining useful life (RUL). This paper studies a contrastive adversarial variational recurrent method for machine RUL prediction under different service conditions. In this new approach, the variational recurrent networks are developed to extract distribution features and latent spaces of raw data, meanwhile, the adversarial strategies are applied to enable the learning of domain-invariant features to make the model achieve cross-domain task processing. To supervise the learning process of the model to learn more mutual information between the extracted features and the raw input data, a contrastive loss is also introduced in the proposed method. Sufficient experiments were conducted to verify the feasibility and superiority of the suggested approach, including 12 sets of cross-scenario tests on the turbofan engine dataset C-MAPSS from NASA. Experimental findings confirm that the proposed method performs satisfactorily and competitively even compared to current state-of-the-art works.

RetVes segmentation: A pseudo-labeling and feature knowledge distillation optimization technique for retinal vessel channel enhancement

Recent advancements in retinal vessel segmentation, which employ transformer-based and domain-adaptive approaches, show promise in addressing the complexity of ocular diseases such as diabetic retinopathy. However, current algorithms face challenges in effectively accommodating domain-specific variations and limitations of training datasets, which fail to represent real-world conditions comprehensively. Manual inspection by specialists remains time-consuming despite technological progress in medical imaging, underscoring the pressing need for automated and robust segmentation techniques. Additionally, these methods have deficiencies in handling unlabeled target sets, requiring extra preprocessing steps and manual intervention, which hinders their scalability and practical application in clinical settings. This research introduces a novel framework that employs semi-supervised domain adaptation and contrastive pre-training to address these limitations. The proposed model effectively learns from target data by implementing a novel pseudo-labeling approach and feature-based knowledge distillation within a temporal convolutional network (TCN) and extracts robust, domain-independent features. This approach enhances cross-domain adaptation, significantly enhancing the model's versatility and performance in clinical settings. The semi-supervised domain adaptation component overcomes the challenges posed by domain shifts, while pseudo-labeling utilizes the data's inherent structure for enhanced learning, which is particularly beneficial when labeled data is scarce. Evaluated on the DRIVE and CHASE_DB1 datasets, which contain clinical fundus images, the proposed model achieves outstanding performance, with accuracy, sensitivity, specificity, and AUC values of 0.9792, 0.8640, 0.9901, and 0.9868 on DRIVE, and 0.9830, 0.9058, 0.9888, and 0.9950 on CHASE_DB1, respectively, outperforming current state-of-the-art vessel segmentation methods. The partitioning of datasets into training and testing sets ensures thorough validation, while extensive ablation studies with thorough sensitivity analysis of the model's parameters and different percentages of labeled data further validate its robustness.

The Evaluation of Machine Learning Techniques for Isotope Identification Contextualized by Training and Testing Spectral Similarity

Precise gamma-ray spectral analysis is crucial in high-stakes applications, such as nuclear security. Research efforts toward implementing machine learning (ML) approaches for accurate analysis are limited by the resemblance of the training data to the testing scenarios. The underlying spectral shape of synthetic data may not perfectly reflect measured configurations, and measurement campaigns may be limited by resource constraints. Consequently, ML algorithms for isotope identification must maintain accurate classification performance under domain shifts between the training and testing data. To this end, four different classifiers (Ridge, Random Forest, Extreme Gradient Boosting, and Multilayer Perceptron) were trained on the same dataset and evaluated on twelve other datasets with varying standoff distances, shielding, and background configurations. A tailored statistical approach was introduced to quantify the similarity between the training and testing configurations, which was then related to the predictive performance. Wilcoxon signed-rank tests revealed that the OVR-wrapped XGB significantly outperformed the other algorithms, with confidence levels of 99.0% or above for the 133Ba, 60Co, 137Cs, and 152Eu sources. The findings from this work are significant as they outline techniques to promote the development of robust ML-based approaches for isotope identification.