Search Results (29)

Federated, semi-supervised novelty detection is well suited for intrusion detection on resource-constrained Internet of Things (IoT) nodes: each device learns a model of benign traffic, shares only summary statistics, and does not transmit raw traffic samples. A previously published cross-layer federated detector, Baseline K-Means, showed that periodically merging worker statistics through a coordinator raises the detection rate, but it also exhibited a systematic side effect: after every merge, the precision decays, and the false-positive rate (FPR) climbs because the coordinator recomputes its threshold from streaming distances filtered by the closest observed anomaly, so tightens after every merge, flagging progressively more benign traffic; the threshold was also hand-tuned. We present AF-BKM, an Adaptive Federated Baseline K-Means that repairs the federated mechanism with two label-free, statistics-only enhancements, denoted as E1 and E2: (i) an adaptive decision threshold read from the benign Mahalanobis-distance distribution, requiring no manual percentile search and no attack labels (E1), and (ii) a robust, benignly anchored aggregation that blends worker means under quality weighting and outlier-worker filtering and recalibrates the threshold on a trusted benign anchor to a stable, anchor-referenced false-positive level, which a target-FPR rule can make operator-selectable instead of tightening it toward the nearest anomaly (E2). With MinMax scaling fit only on benign baseline data and non-IID federated streams on NSL-KDD, UNSW-NB15 and the N-BaIoT corpus of real traffic from commercial IoT devices, AF-BKM removes the merge-induced precision decay (the first-to-last-epoch precision change improves from $-0.134$ to $-0.002$ on NSL-KDD, from $-0.121$ to $-0.014$ on UNSW-NB15, and from $-0.170$ to $-0.009$ on N-BaIoT) and reduces the mean FPR by 30–$64\%$, depending on the dataset; all central improvements are significant across 10 seeds (Wilcoxon $p=0.002$, large effect sizes). AF-BKM preserves recall on NSL-KDD and N-BaIoT and, on the harder UNSW-NB15, exposes an explicit precision–recall trade-off through a benign target-FPR knob. In fp32, the deployed model serializes to $5.5$–52 KB, a packet is classified in 11–27 µs on a desktop CPU, and each merge round uploads a $d+3$-value summary (160–472 B) $94.6$–$98.3\%$ smaller than the same summary extended with the covariance upper triangle. A robustness study covering selected faulty-worker updates, contamination of the commissioning anchor, and detector-level white-box evasion reports the measured degradation patterns: fabricated threshold candidates have no direct path to the threshold, although a fabricated mean still reaches it indirectly through the blended centroid, and the anchor-referenced false-positive level remains stable under percent-level anchor contamination, while recall sensitivity is dataset-dependent and the evasion budget tracks the benign–attack margin of each dataset. We frame the contribution with a focused taxonomy that identifies merge-induced precision decay under non-IID workers as an open gap. Code is released for reproducibility. Full article

Segmenting key components of space targets using Inverse Synthetic Aperture Radar (ISAR) images is an important interpretation task in space situational awareness. However, the scarcity of pixel-level annotated data, inter-class confusion caused by morphological differences among multiple target classes, and the absence of structural priors for components restrict the performance improvement in existing deep models on this task. Therefore, this paper proposes a Semi-Supervised Structural Prior-Guided Network (SSPNet). First, a Gated Manifold-Constrained Hyper-Connections Vision Transformer (GMHC-ViT) encoder is proposed to broaden the feature representation space via parallel multi-feature streams with adaptive gating, thereby alleviating inter-class confusion and enhancing cross-category generalization. Second, a Prior-Guided Module (PGM) is proposed to extract shape and edge priors of components, and it adaptively enhances the weakly activated channels of encoder features through cross-attention, thereby injecting structural knowledge independent of image quality into the segmentation process. Furthermore, to effectively leverage large amounts of unlabeled data, a strong perturbation strategy tailored to the characteristics of ISAR images is designed for consistency regularization. Experimental results on a simulated ISAR dataset containing 38 classes of space targets demonstrate that SSPNet outperforms existing methods and exhibits strong segmentation capability even under low signal-to-noise ratio (SNR) conditions. Full article

Background/Objectives: Chronic kidney disease is a growing global health concern, necessitating effective tools for early detection and monitoring. While non-contrast T1-weighted magnetic resonance imaging offers a non-invasive means to assess kidney morphology, robust automated segmentation remains challenging due to limited annotated data, high inter-patient variability, and low signal-to-noise ratios. Methods: In this study, we address these obstacles by developing and evaluating a series of frugal learning methodologies for kidney segmentation in non-contrast MRI. Building upon the U-Net architecture, we aim to maximize segmentation accuracy despite scarce labeled data. Our experimental framework leverages three diverse datasets to evaluate performance-boosting strategies such as transfer learning: a clinically relevant local cohort (the Barlicki dataset) as the primary target domain and two auxiliary public datasets (AMOS22 and AbdomenCT). Utilizing these data streams, we systematically compare seven frugal learning strategies incorporating data augmentation, semi-supervised learning, and weak supervision against a fully supervised baseline. Results: The results demonstrate that frugal learning methods enable accurate and reliable kidney segmentation while substantially reducing the need for manual annotations. The best-performing semi-supervised and transfer learning approaches achieved a Dice similarity coefficient of 0.89, which was only moderately lower than that of the fully supervised model (Dice = 0.92). Conclusions: This work highlights the potential of data-efficient deep learning techniques to accelerate the adoption of automated kidney segmentation in clinical workflows, particularly in settings where annotated medical images are limited. Full article

Deep learning has revolutionized medical image segmentation; however, the clinical deployment of state-of-the-art models is severely impeded by their quadratic computational complexity and substantial resource demands, particularly in multisensor and multimodal imaging scenarios. In response, the field is undergoing a paradigm shift towards efficiency, characterized by the rise of linear-complexity architectures and the optimization of foundation models. This paper presents a comprehensive survey of efficient medical image segmentation methodologies, systematically reviewing the evolution from heavy, accuracy-driven models to lightweight, deployment-ready paradigms. In particular, we highlight the growing importance of efficient segmentation in multisensor medical imaging, where heterogeneous data sources such as CT, MRI, ultrasound, and infrared imaging introduce additional challenges in scalability and computational cost. We propose a novel taxonomy that categorizes these advancements into four distinct streams: (1) Mamba and State Space Models, which leverage selective scanning mechanisms to achieve global receptive fields with linear complexity; (2) Efficient Adaptation of Foundation Models, focusing on parameter-efficient fine-tuning and knowledge distillation to tailor the Segment Anything Model (SAM) for medical domains; (3) Advanced Lightweight Architectures, covering the resurgence of large-kernel CNNs and the emergence of Kolmogorov–Arnold Networks (KANs); and (4) Data-Efficient Strategies, including semi-supervised and federated learning to address annotation scarcity. Furthermore, we conduct a rigorous comparative analysis of representative algorithms on mainstream benchmarks, providing a granular evaluation of the trade-offs between segmentation accuracy and computational overhead. The survey also discusses key challenges in multisensor and multimodal settings, including modality heterogeneity, data fusion complexity, and resource constraints. Finally, we identify critical challenges and outline future research directions, serving as a roadmap for the development of next-generation efficient and scalable medical image analysis systems. Full article

In sensor-centric fields like healthcare, environmental monitoring, and industry, image classification is key to turning visual sensor data into actionable insights. Sensor-generated dynamic streaming data poses significant challenges for traditional static image classification models due to the continuous emergence of new categories, distribution shifts, and limited edge storage. With sensor streaming data facing label scarcity and high annotation costs, semi-supervised continual learning is essential, leveraging unlabeled data for incremental learning and reducing reliance on costly annotations. However, current semi-supervised continual learning methods rely on labeled data to generate pseudo-labels, leading to confirmation and relational biases. To mitigate these dual biases, we propose a Bias Calibration method based on nearest-neighbor semi-supervised continual learning, which integrates and adapts Confidence-Enhanced Learning (originally introduced for static datasets) and Guided Contrastive Learning. Specifically, the Confidence-Enhanced Learning aims to reduce competition among similar classes and penalizes low-confidence predictions, thereby generating high-confidence pseudo-labels for unlabeled data and mitigating confirmation bias. Guided Contrastive Learning constructs a pseudo-label graph and a feature representation graph, using the pseudo-label graph to optimize the feature representation graph, thereby improving class discrimination and reducing feature bias. Experiments on CIFAR-10, CIFAR-100, and ImageNet-100 show that our method significantly outperforms existing approaches, enhancing classification performance with partial labeling. Full article

Conventional Security Operations Center (SOC) solutions struggle to process representative operational alert streams efficiently and adapt to evolving cyber threats, highlighting the need for automated, intelligent threat detection and prioritization. This study presents a custom AI-driven framework that leverages unsupervised learning techniques to support SOC analysts in cyber threat detection, anomaly identification, and alert prioritization. The framework applies several clustering methods: HDBSCAN, DBSCAN, KMeans, and Gaussian Mixture Models for alert segmentation, and integrates anomaly detection using LOF and Isolation Forest, complemented by semi-supervised detection via One-Class SVM. Using textual, categorical, and numerical features from Wazuh alerts across three datasets, the system performs clustering and anomaly detection in the original high-dimensional feature space, with UMAP applied solely for two-dimensional visualization. HDBSCAN consistently produced well-separated clusters with effective noise detection, while, Isolation Forest evaluated via 10-fold cross-validation exhibited stable anomaly flagging and clear score separation across both cyber alert event data and synthetic threat injection experiments. Furthermore, the framework formulates a composite priority ranking that integrates anomaly severity, cluster rarity, and SOC contextual weighting, yielding actionable alert rankings. An interactive, analyst-centric dashboard enables SOC teams to explore top alerts, clusters, associated MITRE techniques, priority rankings, and geolocation data, providing insights while preserving human oversight. Overall, the proposed system transforms complex alert streams into structured insights, enhancing SOC situational awareness, decision support, and operational efficiency. Full article

Anomaly detection in safety-critical systems often operates under severe label constraints, where only a small subset of normal and anomalous samples can be reliably annotated, while large unlabeled data streams are contaminated and high-dimensional. Deep one-class methods, such as deep support vector data description (DeepSVDD) and deep semi-supervised anomaly detection (DeepSAD), address this setting. However, they treat samples largely in isolation and do not explicitly leverage the manifold structure of unlabeled data, which can limit robustness and interpretability. This paper proposes Anomaly-Aware Graph-based Semi-Supervised Deep Support Vector Data Description (AAG-DSVDD), a boundary-focused deep one-class approach that couples a DeepSAD-style hypersphere with a label-aware latent k-nearest neighbor (k-NN) graph. The method combines a soft-boundary enclosure for labeled normals, a margin-based push-out for labeled anomalies, an unlabeled center-pull, and a k-NN graph regularizer on the squared distances to the center. The resulting graph term propagates information from scarce labels along the latent manifold, aligns anomaly scores of neighboring samples, and supports sample-level interpretability through graph neighborhoods, while test-time scoring remains a single distance-to-center computation. On a controlled two-dimensional synthetic dataset, AAG-DSVDD achieves a mean F1-score of $0.88\pm 0.02$ across ten random splits, improving on the strongest baseline by about $0.12$ absolute F1. On three public benchmark datasets (Thyroid, Arrhythmia, and Heart), AAG-DSVDD attains the highest F1 on all datasets with F1-scores of $0.719$, $0.675$, and $0.8$, respectively, compared to all baselines. In a multi-sensor fire monitoring case study, AAG-DSVDD reduces the average absolute error in fire starting time to approximately 473 s (about 30% improvement over DeepSAD) while keeping the average pre-fire false-alarm rate below $1\%$ and avoiding persistent pre-fire alarms. These results indicate that graph-regularized deep one-class boundaries offer an effective and interpretable framework for semi-supervised anomaly detection under realistic label budgets. Full article

We present a novel deep generative semi-supervised framework for credit card fraud detection, formulated as a time series classification task. As financial transaction data streams grow in scale and complexity, traditional methods often require large labeled datasets and struggle with time series of irregular sampling frequencies and varying sequence lengths. To address these challenges, we extend conditional Generative Adversarial Networks (GANs) for targeted data augmentation, integrate Bayesian inference to obtain predictive distributions and quantify uncertainty, and leverage log-signatures for robust feature encoding of transaction histories. We propose a composite Wasserstein distance-based loss to align generated and real unlabeled samples while simultaneously maximizing classification accuracy on labeled data. Our approach is evaluated on the BankSim dataset, a widely used simulator for credit card transaction data, under varying proportions of labeled samples, demonstrating consistent improvements over benchmarks in both global statistical and domain-specific metrics. These findings highlight the effectiveness of GAN-driven semi-supervised learning with log-signatures for irregularly sampled time series and emphasize the importance of uncertainty-aware predictions. Full article

Atmospheric jets are pivotal components of atmospheric circulation, profoundly influencing surface weather patterns and the development of extreme weather events such as storms and cold waves. Accurate detection of the jet stream axis is indispensable for enhancing weather forecasting, monitoring climate change, and mitigating disasters. However, traditional methods for delineating atmospheric jets are plagued by inefficiency, substantial errors, and pronounced subjectivity, limiting their applicability in complex atmospheric scenarios. Current research on semi-supervised methods for extracting atmospheric jets remains scarce, with most approaches dependent on traditional techniques that struggle with stability and generalization. To address these limitations, this study proposes a semi-supervised jet stream axis extraction method leveraging an enhanced U-Net++ model. The approach incorporates improved residual blocks and enhanced attention gate mechanisms, seamlessly integrating these enhanced attention gates into the dense skip connections of U-Net++. Furthermore, it optimizes the consistency learning phase within semi-supervised frameworks, effectively addressing data scarcity challenges while significantly enhancing the precision of jet stream axis detection. Experimental results reveal the following: (1) With only 30% of labeled data, the proposed method achieves a precision exceeding 80% on the test set, surpassing state-of-the-art (SOTA) baselines. Compared to fully supervised U-Net and U-Net++ methods, the precision improves by 17.02% and 9.91%. (2) With labeled data proportions of 10%, 20%, and 30%, the proposed method outperforms the MT semi-supervised method, achieving precision gains of 9.44%, 15.58%, and 19.50%, while surpassing the DCT semi-supervised method with improvements of 10.24%, 16.64%, and 14.15%, respectively. Ablation studies further validate the effectiveness of the proposed method in accurately identifying the jet stream axis. The proposed method exhibits remarkable consistency, stability, and generalization capabilities, producing jet stream axis extractions closely aligned with wind field data. Full article

The scarcity of high-quality labeled data poses a challenge to the application of deep learning (DL) in landslide identification from remote sensing (RS) images. Semi-supervised learning (SSL) has emerged as a promising approach to address the issue of low accuracy caused by the limited availability of high-quality labels. Nevertheless, the application of SSL approaches developed for natural images to landslide identification encounters several challenges. This study focuses on two specific challenges: inadequate information extraction from limited unlabeled RS landslide images and the generation of low-quality pseudo-labels. To tackle these challenges, we propose a novel and generic DL framework called hybrid perturbation mean match (HPM-Match). The framework combines dual-branch input perturbation (DIP) and independent triple-stream perturbation (ITP) techniques to enhance model accuracy with limited labels. The DIP generation approach is designed to maximize the utilization of manually pre-defined perturbation spaces while minimizing the introduction of erroneous information during the weak-to-strong consistency learning (WSCL) process. Moreover, the ITP structure unifies input, feature, and model perturbations, thereby broadening the perturbation space and enabling knowledge extraction from unlabeled landslide images across various perspectives. Experimental results demonstrate that HPM-Match has substantial improvements in IoU, with maximum increases of 26.68%, 7.05%, and 12.96% over supervised learning across three datasets with the same label ratio and reduces the number of labels by up to about 70%. Furthermore, HPM-Match strikes a better balance between precision and recall, identifying more landslides than other state-of-the-art (SOTA) SSL approaches. Full article

Semi-supervised medical image segmentation models often face challenges such as empirical mismatch and data imbalance. Traditional methods, like the two-stream perturbation model, tend to over-rely on strong perturbation, leaving weak perturbation and labeled images underutilized. To overcome these challenges, we propose an innovative hybrid copy-paste (HCP) method within the strong perturbation branch, encouraging unlabeled images to learn more comprehensive semantic information from labeled images and narrowing the empirical distribution gap. Additionally, we integrate contrastive learning into the weak perturbation branch, where contrastive learning samples are selected through semantic grouping contrastive sampling (SGCS) to address memory and variance issues. This sampling strategy ensures more effective use of weak perturbation data. This approach is particularly advantageous for pixel segmentation tasks with severely limited labels. Finally, our approach is validated on the public ACDC (Automated Cardiac Diagnosis Challenge) dataset, achieving a 90.6% DICE score, with just 7% labeled data. These results demonstrate the effectiveness of our method in improving segmentation performance with limited labeled data. Full article

Changes in the jet stream not only affect the persistence of climate change and the frequency of extreme weather but are also closely related to climate change phenomena such as global warming. The manual way of drawing the jet stream axes in meteorological operations suffers from low efficiency and subjectivity issues. Automatic identification algorithms based on wind field analysis have some shortcomings, such as poor generalization ability, and it is difficult to handle merging and splitting. A semi-supervised learning jet stream axis identification method is proposed combining consistency learning and self-training. First, a segmentation model is trained via semi-supervised learning. In semi-supervised learning, two neural networks with the same structure are initialized with different methods, based on which pseudo-labels are obtained. The high-confidence pseudo-labels are selected by adding perturbation into the feature layer, and the selected pseudo-labels are incorporated into the training set for further self-training. Then, the jet stream narrow regions are segmented via the trained segmentation model. Finally, the jet stream axes are obtained with the skeleton extraction method. This paper uses the semi-supervised jet stream axis identification method to learn features from unlabeled data to achieve a small amount of labeled data to effectively train the model and improve the method’s generalization ability in a small number of labeled cases. Experiments on the jet stream axis dataset show that the identification precision of the presented method on the test set exceeds about 78% for SOTA baselines, and the improved method exhibits better performance compared to the correlation network model and the semi-supervised method. Full article

This paper describes a multistage framework for face image analysis in computer-aided speech diagnosis and therapy. Multimodal data processing frameworks have become a significant factor in supporting speech disorders’ treatment. Synchronous and asynchronous remote speech therapy approaches can use audio and video analysis of articulation to deliver robust indicators of disordered speech. Accurate segmentation of articulators in video frames is a vital step in this agenda. We use a dedicated data acquisition system to capture the stereovision stream during speech therapy examination in children. Our goal is to detect and accurately segment four objects in the mouth area (lips, teeth, tongue, and whole mouth) during relaxed speech and speech therapy exercises. Our database contains 17,913 frames from 76 preschool children. We apply a sequence of procedures employing artificial intelligence. For detection, we train the YOLOv6 (you only look once) model to catch each of the three objects under consideration. Then, we prepare the DeepLab v3+ segmentation model in a semi-supervised training mode. As preparation of reliable expert annotations is exhausting in video labeling, we first train the network using weak labels produced by initial segmentation based on the distance-regularized level set evolution over fuzzified images. Next, we fine-tune the model using a portion of manual ground-truth delineations. Each stage is thoroughly assessed using the independent test subset. The lips are detected almost perfectly (average precision and F1 score of 0.999), whereas the segmentation Dice index exceeds 0.83 in each articulator, with a top result of 0.95 in the whole mouth. Full article

Continual learning (CL) from infinite data streams has become a challenge for neural network models in real-world scenarios. Catastrophic forgetting of previous knowledge occurs in this learning setting, and existing supervised CL methods rely excessively on accurately labeled samples. However, the real-world data labels are usually misled by noise, which influences the CL agents and aggravates forgetting. To address this problem, we propose a method named noise-labeled online continual learning (NLOCL), which implements the online CL model with noise-labeled data streams. NLOCL uses an empirical replay strategy to retain crucial examples, separates data streams by small-loss criteria, and includes semi-supervised fine-tuning for labeled and unlabeled samples. Besides, NLOCL combines small loss with class diversity measures and eliminates online memory partitioning. Furthermore, we optimized the experience replay stage to enhance the model performance by retaining significant clean-labeled examples and carefully selecting suitable samples. In the experiment, we designed noise-labeled data streams by injecting noisy labels into multiple datasets and partitioning tasks to simulate infinite data streams realistically. The experimental results demonstrate the superior performance and robust learning capabilities of our proposed method. Full article

Anomaly detection in data streams (and particularly time series) is today a vitally important task. Machine learning algorithms are a common design for achieving this goal. In particular, deep learning has, in the last decade, proven to be substantially more accurate than shallow learning in a wide variety of machine learning problems, and deep anomaly detection is very effective for point anomalies. However, deep semi-supervised contextual anomaly detection (in which anomalies within a time series are rare and none at all occur in the algorithm’s training data) is a more difficult problem. Hybrid anomaly detectors (a “normal model” followed by a comparator) are one approach to these problems, but the separate loss functions for the two components can lead to inferior performance. We investigate a novel synthetic-example oversampling technique to harmonize the two components of a hybrid system, thus improving the anomaly detector’s performance. We evaluate our algorithm on two distinct problems: identifying pipeline leaks and patient-ventilator asynchrony. Full article