Search Results (61)

The medical imaging domain frequently encounters the dual challenges of annotation scarcity and class imbalance. A critical issue lies in effectively extracting information from limited labeled data while mitigating the dominance of head classes. The existing approaches often overlook in-depth modeling of sample relationships in low-dimensional spaces, while rigid or suboptimal dynamic thresholding strategies in pseudo-label generation are susceptible to noisy label interference, leading to cumulative bias amplification during the early training phases. To address these issues, we propose a semi-supervised medical image classification framework combining labeled data-contrastive learning with class-adaptive pseudo-labeling (CLCP-MT), comprising two key components: the semantic discrimination enhancement (SDE) module and the class-adaptive pseudo-label refinement (CAPR) module. The former incorporates supervised contrastive learning on limited labeled data to fully exploit discriminative information in latent structural spaces, thereby significantly amplifying the value of sparse annotations. The latter dynamically calibrates pseudo-label confidence thresholds according to real-time learning progress across different classes, effectively reducing head-class dominance while enhancing tail-class recognition performance. These synergistic modules collectively achieve breakthroughs in both information utilization efficiency and model robustness, demonstrating superior performance in class-imbalanced scenarios. Extensive experiments on the ISIC2018 skin lesion dataset and Chest X-ray14 thoracic disease dataset validate CLCP-MT’s efficacy. With only 20% labeled and 80% unlabeled data, our framework achieves a 10.38% F1-score improvement on ISIC2018 and a 2.64% AUC increase on Chest X-ray14 compared to the baselines, confirming its effectiveness and superiority under annotation-deficient and class-imbalanced conditions. Full article

The spread of misinformation during the COVID-19 pandemic raised widespread concerns about public health communication and media reliability. In this study, we focus on these issues as they manifested in Romanian-language media and employ Large Language Models (LLMs) to classify misinformation, with a particular focus on super-narratives—broad thematic categories that capture recurring patterns and ideological framings commonly found in pandemic-related fake news, such as anti-vaccination discourse, conspiracy theories, or geopolitical blame. While some of the categories reflect global trends, others are shaped by the Romanian cultural and political context. We introduce a novel dataset of fake news centered on COVID-19 misinformation in the Romanian geopolitical context, comprising both annotated and unannotated articles. We experimented with multiple LLMs using zero-shot, few-shot, supervised, and semi-supervised learning strategies, achieving the best results with an LLaMA 3.1 8B model and semi-supervised learning, which yielded an F1-score of 78.81%. Experimental evaluations compared this approach to traditional Machine Learning classifiers augmented with morphosyntactic features. Results show that semi-supervised learning substantially improved classification results in both binary and multi-class settings. Our findings highlight the effectiveness of semi-supervised adaptation in low-resource, domain-specific contexts, as well as the necessity of enabling real-time misinformation tracking and enhancing transparency through claim-level explainability and fact-based counterarguments. Full article

Underwater acoustic target recognition (UATR) is challenged by complex marine noise, scarce labeled data, and inadequate multi-scale feature extraction in conventional methods. This study proposes DART-MT, a semi-supervised framework that integrates a Dual Attention Parallel Residual Network Transformer with a mean teacher paradigm, enhanced by domain-specific prior knowledge. The architecture employs a Convolutional Block Attention Module (CBAM) for localized feature refinement, a lightweight New Transformer Encoder for global context modeling, and a novel TriFusion Block to synergize spectral–temporal–spatial features through parallel multi-branch fusion, addressing the limitations of single-modality extraction. Leveraging the mean teacher framework, DART-MT optimizes consistency regularization to exploit unlabeled data, effectively mitigating class imbalance and annotation scarcity. Evaluations on the DeepShip and ShipsEar datasets demonstrate state-of-the-art accuracy: with 10% labeled data, DART-MT achieves 96.20% (DeepShip) and 94.86% (ShipsEar), surpassing baseline models by 7.2–9.8% in low-data regimes, while reaching 98.80% (DeepShip) and 98.85% (ShipsEar) with 90% labeled data. Under varying noise conditions (−20 dB to 20 dB), the model maintained a robust performance (F1-score: 92.4–97.1%) with 40% lower variance than its competitors, and ablation studies validated each module’s contribution (TriFusion Block alone improved accuracy by 6.9%). This research advances UATR by (1) resolving multi-scale feature fusion bottlenecks, (2) demonstrating the efficacy of semi-supervised learning in marine acoustics, and (3) providing an open-source implementation for reproducibility. In future work, we will extend cross-domain adaptation to diverse oceanic environments. Full article

The intelligent development of continuous casting quality engineering is an essential step for the efficient production of high-quality billets. However, there are many quality defects that require strong expertise for handling. In order to reduce reliance on expert experience and improve the intelligent management level of billet quality knowledge, we focus on constructing a Domain-Specific Knowledge Graph (DSKG) for the quality engineering of continuous casting. To achieve joint extraction of billet quality defects entity and relation, we propose a Self-Attention Partition and Recombination Model (SAPRM). SAPRM divides domain-specific sentences into three parts: entity-related, relation-related, and shared features, which are specifically for Named Entity Recognition (NER) and Relation Extraction (RE) tasks. Furthermore, for issues of entity ambiguity and repetition in triples, we propose a semi-supervised incremental learning method for knowledge alignment, where we leverage adversarial training to enhance the performance of knowledge alignment. In the experiment, in the knowledge extraction part, the NER and RE precision of our model achieved 86.7% and 79.48%, respectively. RE precision improved by 20.83% compared to the baseline with sequence labeling method. Additionally, in the knowledge alignment part, the precision of our model reached 99.29%, representing a 1.42% improvement over baseline methods. Consequently, the proposed model with the partition mechanism can effectively extract domain knowledge, cand the semi-supervised method can take advantage of unlabeled triples. Our method can adapt the domain features and construct a high-quality knowledge graph for the quality engineering of continuous casting, providing an efficient solution for billet defect issues. Full article

Surface-related multiple suppression is a critical step in seismic data processing, while traditional adaptive matching subtraction methods often distort primaries, resulting in either the leakage of primaries or the residue of surface-related multiples. To address these challenges, we propose a field-parameter-guided semi-supervised learning (FPSSL) method to more effectively eliminate surface-related multiples. Field parameters refer to the time–space coordinate information derived from the seismic acquisition system, including offsets, trace spaces, and sampling intervals. These parameters reveal the relative positional relationships of seismic data in the time–space domain. The FPSSL framework comprises a supervised network module (SNM) and an unsupervised network module (USNM). The input and output data of the SNM are a small sample of full wavefield data and the weights of a polynomial function, respectively. A linear weighted sum method is employed to represent the SNM outputs (weights), the full wavefield data, and field parameters as a polynomial function of the primaries, which is matched with adaptive subtraction label data. The trained SNM generates preliminary estimates of the primaries and multiples with improved lateral continuity from full wavefield data, both of which are used as inputs to the USNM. The USNM is essentially an optimization operator that refines the underlying nonlinear mapping relationship between primaries and full wavefield data using the local wavefield feature loss function, thereby obtaining more accurate prediction results with respect to primaries. Examples from synthetic data and real marine data demonstrate that the FPSSL method surpasses the traditional L1-norm adaptive subtraction method in suppressing multiples, significantly reducing the leakage of primaries and the residuals of surface-related multiples in the estimated demultiple results. The effectiveness and efficiency of our proposed method are verified through two sets of synthetic data and one marine data example. Full article

In open-pit mining, autonomous trucks are essential for enhancing both safety and productivity. Object detection technology is critical to their smooth and secure operation, but training these models requires large amounts of high-quality annotated data representing various conditions. It is expensive and time-consuming to collect these data during open-pit mining due to the harsh environmental conditions. Simulation engines have emerged as an effective alternative, generating diverse labeled data to augment real-world datasets. However, discrepancies between simulated and real-world environments, often referred to as the Sim2Real domain shift, reduce model performance. This study addresses these challenges by presenting a novel semi-supervised domain adaptation for object detection (SSDA-OD) framework named Adamix, which is designed to reduce domain shift, enhance object detection, and minimize labeling costs. Adamix builds on a mean teacher architecture and introduces two key modules: progressive intermediate domain construction (PIDC) and warm-start adaptive pseudo-label (WSAPL). PIDC builds intermediate domains using a mixup strategy to reduce source domain bias and prevent overfitting, while WSAPL provides adaptive thresholds for pseudo-labeling, mitigating false and missed detections during training. When evaluated in a Sim2Real scenario, Adamix shows superior domain adaptation performance, achieving a higher mean average precision (mAP) compared with state-of-the-art methods, with 50% less labeled data required, achieved through active learning. The results demonstrate that Adamix significantly reduces dependence on costly real-world data collection, offering a more efficient solution for object detection in challenging open-pit mining environments. Full article

Due to the challenge of acquiring abundant labeled samples, semi-supervised change detection (SSCD) approaches are becoming increasingly popular in tackling CD tasks with limited labeled data. Despite their success, these methods tend to come with complex network architectures or cumbersome training procedures, which also ignore the domain gap between the labeled data and unlabeled data. Differently, we hypothesize that diverse perturbations are more favorable to exploit the potential of unlabeled data. In light of this spirit, we propose a novel SSCD approach based on Weak–strong Augmentation and Class-balanced Sampling (WACS-SemiCD). Specifically, we adopt a simple mean-teacher architecture to deal with labeled branch and unlabeled branch separately, where supervised learning is conducted on the labeled branch, while weak–strong consistency learning (e.g., sample perturbations’ consistency and feature perturbations’ consistency) is imposed for the unlabeled. To improve domain generalization capacity, an adaptive CutMix augmentation is proposed to inject the knowledge from the labeled data into the unlabeled data. A class-balanced sampling strategy is further introduced to mitigate class imbalance issues in CD. Particularly, our proposed WACS-SemiCD achieves competitive SSCD performance on three publicly available CD datasets under different labeled settings. Comprehensive experimental results and systematic analysis underscore the advantages and effectiveness of our proposed WACS-SemiCD. Full article

Heterogeneous domain adaptation (HDA) utilizes the knowledge of the source domain to model the target domain. Although the two domains are semantically related, the problem of feature and distribution differences in heterogeneous data still needs to be solved. Most of the existing HDA methods only consider the feature or distribution problem but do not consider the geometric semantic information similarity between the domain structures, which leads to a weakened adaptive performance. In order to solve the problem, a centroid connected structure matching network (CCSMN) approach is proposed, which firstly maps the heterogeneous data into a shared public feature subspace to solve the problem of feature differences. Secondly, it promotes the overlap of domain centers and nodes of the same category between domains to reduce the positional distribution differences in the internal structure of data. Then, the supervised information is utilized to generate target domain nodes, and the geometric structural and semantic information are utilized to construct a centroid-connected structure with a reasonable inter-class distance. During the training process, a progressive and integrated pseudo-labeling is utilized to select samples with high-confidence labels and improve the classification accuracy for the target domain. Extensive experiments are conducted in text-to-image and image-to-image HDA tasks, and the results show that the CCSMN outperforms several state-of-the-art baseline methods. Compared with state-of-the-art HDA methods, in the text-to-image transfer task, the efficiency has increased by 8.05%; and in the image-to-image transfer task, the efficiency has increased by about 2%, which suggests that the CCSMN benefits more from domain geometric semantic information similarity. Full article

As an effective computer-aided diagnostic tool, deep learning has been successfully applied to the classification of breast histopathological images. However, the performance of the deep model is data-driven, and it is difficult to obtain satisfied results when the number of histopathological images is small and labelling histopathological images is difficult. Moreover, in traditional deep learning methods, the representation of features is monotonous, which leads to the limitation of the classification performance of the model. This study proposes an auto-encoder reconstructed semi-supervised domain adaptation for a breast histopathological image classification algorithm. First, the model was pre-trained and transferred to extract high-level features of the sample images. Then, the encoding and decoding parts of the auto-encoder were used to reconstruct the feature representation learning and the sample feature reconstruction learning, respectively. This ensured that the useful information for the classification was purified and retained. At the same time, the domain discriminator was used to confuse the source and target domain features to enhance the learning ability of the model. Finally, the distribution difference of features at different depths of the auto-encoder was measured to minimize the discrepancy of feature distribution between domains, so as to complete the classification of histopathological images. Compared to the results of the comparative and ablation algorithms from the BreakHis to SNL datasets, the proposed method achieved the best results in terms of F1 score (93.40%), accuracy (95.24%), sensitivity (94.66%), and specificity (95.56%). The experimental results demonstrate that the proposed method achieves a remarkable classification performance. Full article

Background: Advances in medical image classification have recently benefited from general augmentation techniques. However, these methods often fall short in performance and interpretability. Objective: This paper applies the Unified Model Agnostic Computation (UMAC) framework specifically to the medical domain to demonstrate its utility in this critical area. Methods: UMAC is a model-agnostic methodology designed to develop machine learning approaches that integrate seamlessly with various paradigms, including self-supervised, semi-supervised, and supervised learning. By unifying and standardizing computational models and algorithms, UMAC ensures adaptability across different data types and computational environments while incorporating state-of-the-art methodologies. In this study, we integrate UMAC as a plug-and-play module within convolutional neural networks (CNNs) and Transformer architectures, enabling the generation of high-quality representations even with minimal data. Results: Our experiments across nine diverse 2D medical image datasets show that UMAC consistently outperforms traditional data augmentation methods, achieving a 1.89% improvement in classification accuracy. Conclusions: Additionally, by incorporating explainable AI (XAI) techniques, we enhance model transparency and reliability in decision-making. This study highlights UMAC’s potential as a powerful tool for improving both the performance and interpretability of medical image classification models. Full article

Existing dehazing methods deal with real-world haze images with difficulty, especially scenes with thick haze. One of the main reasons is lacking real-world pair data and robust priors. To improve dehazing ability in real-world scenes, we propose a semi-supervised codebook learning dehazing method. The codebook is used as a strong prior to guide the hazy image recovery process. However, the following two issues arise when the codebook is applied to the image dehazing task: (1) Latent space features obtained from the coding of degraded hazy images suffer from matching errors when nearest-neighbour matching is performed. (2) Maintaining a good balance of image recovery quality and fidelity for heavily degraded dense hazy images is difficult. To reduce the nearest-neighbor matching error rate in the vector quantization stage of VQGAN, we designed the unit dual-attention residual transformer module (UDART) to correct the latent space features. The UDART can make the latent features obtained from the encoding stage closer to those of the corresponding clear image. To balance the quality and fidelity of the dehazing result, we design a haze density guided weight adaptive module (HDGWA), which can adaptively adjust the multi-scale skip connection weights according to haze density. In addition, we use mean teacher, a semi-supervised learning strategy, to bridge the domain gap between synthetic and real-world data and enhance the model generalization in real-world scenes. Comparative experiments show that our method achieves improvements of 0.003, 2.646, and 0.019 over the second-best method for the no-reference metrics FADE, MUSIQ, and DBCNN, respectively, on the real-world dataset URHI. Full article

Person re-identification (ReID) refers to the task of retrieving target persons from image libraries captured by various distinct cameras. Over the years, person ReID has yielded favorable recognition outcomes under typical visible light conditions, yet there remains considerable scope for enhancement in challenging conditions. The challenges and research gaps include the following: multi-modal data fusion, semi-supervised and unsupervised learning, domain adaptation, ReID in 3D space, fast ReID, decentralized learning, and end-to-end systems. The main problems to be solved, which are the occlusion problem, viewpoint problem, illumination problem, background problem, resolution problem, openness problem, etc., remain challenges. For the first time, this paper uses person ReID in special scenarios as a basis for classification to categorize and analyze the related research in recent years. Starting from the perspectives of person ReID methods and research directions, we explore the current research status in special scenarios. In addition, this work conducts a detailed experimental comparison of person ReID methods employing deep learning, encompassing both system development and comparative methodologies. In addition, we offer a prospective analysis of forthcoming research approaches in person ReID and address unresolved concerns within the field. Full article

Semantic segmentation with cross-domain adaptation in remote-sensing images (RSIs) is crucial and mitigates the expense of manually labeling target data. However, the performance of existing unsupervised domain adaptation (UDA) methods is still significantly impacted by domain bias, leading to a considerable gap compared to supervised trained models. To address this, our work focuses on semi-supervised domain adaptation, selecting a small subset of target annotations through active learning (AL) that maximize information to improve domain adaptation. Overall, we propose a novel active bidirectional self-training network (ABSNet) for cross-domain semantic segmentation in RSIs. ABSNet consists of two sub-stages: a multi-prototype active region selection (MARS) stage and a source-weighted class-balanced self-training (SCBS) stage. The MARS approach captures the diversity in labeled source data by introducing multi-prototype density estimation based on Gaussian mixture models. We then measure inter-domain similarity to select complementary and representative target samples. Through fine-tuning with the selected active samples, we propose an enhanced self-training strategy SCBS, designed for weighted training on source data, aiming to avoid the negative effects of interfering samples. We conduct extensive experiments on the LoveDA and ISPRS datasets to validate the superiority of our method over existing state-of-the-art domain-adaptive semantic segmentation methods. Full article

Precision agriculture involves the application of advanced technologies to improve agricultural productivity, efficiency, and profitability while minimizing waste and environmental impacts. Deep learning approaches enable automated decision-making for many visual tasks. However, in the agricultural domain, variability in growth stages and environmental conditions, such as weather and lighting, presents significant challenges to developing deep-learning-based techniques that generalize across different conditions. The resource-intensive nature of creating extensive annotated datasets that capture these variabilities further hinders the widespread adoption of these approaches. To tackle these issues, we introduce a semi-self-supervised domain adaptation technique based on deep convolutional neural networks with a probabilistic diffusion process, requiring minimal manual data annotation. Using only three manually annotated images and a selection of video clips from wheat fields, we generated a large-scale computationally annotated dataset of image–mask pairs and a large dataset of unannotated images extracted from video frames. We developed a two-branch convolutional encoder–decoder model architecture that uses both synthesized image–mask pairs and unannotated images, enabling effective adaptation to real images. The proposed model achieved a Dice score of 80.7% on an internal test dataset and a Dice score of 64.8% on an external test set composed of images from five countries and spanning 18 domains, indicating its potential to develop generalizable solutions that could encourage the wider adoption of advanced technologies in agriculture. Full article

Ameloblastoma (AM), periapical cyst (PC), and chronic suppurative osteomyelitis (CSO) are prevalent maxillofacial diseases with similar imaging characteristics but different treatments, thus making preoperative differential diagnosis crucial. Existing deep learning methods for diagnosis often require manual delineation in tagging the regions of interest (ROIs), which triggers some challenges in practical application. We propose a new model of Wavelet Extraction and Fusion Module with Vision Transformer (WaveletFusion-ViT) for automatic diagnosis using CBCT panoramic images. In this study, 539 samples containing healthy (n = 154), AM (n = 181), PC (n = 102), and CSO (n = 102) were acquired by CBCT for classification, with an additional 2000 healthy samples for pre-training the domain-adaptive network (DAN). The WaveletFusion-ViT model was initialized with pre-trained weights obtained from the DAN and further trained using semi-supervised learning (SSL) methods. After five-fold cross-validation, the model achieved average sensitivity, specificity, accuracy, and AUC scores of 79.60%, 94.48%, 91.47%, and 0.942, respectively. Remarkably, our method achieved 91.47% accuracy using less than 20% labeled samples, surpassing the fully supervised approach’s accuracy of 89.05%. Despite these promising results, this study’s limitations include a low number of CSO cases and a relatively lower accuracy for this condition, which should be addressed in future research. This research is regarded as an innovative approach as it deviates from the fully supervised learning paradigm typically employed in previous studies. The WaveletFusion-ViT model effectively combines SSL methods to effectively diagnose three types of CBCT panoramic images using only a small portion of labeled data. Full article