Search Results (99)

Soybean seed variety identification is essential for seed quality control, germplasm management, and variety authentication. However, few-shot classification remains challenging because different varieties often exhibit highly similar one-dimensional hyperspectral signatures, and labeled samples are limited in practical seed-testing scenarios. This study proposes a Residual Derivative-Guided Spectral Fusion (RDSF) module to improve spectral representation under limited-sample conditions. RDSF uses the raw spectrum and its first- and second-order derivatives to characterize global reflectance patterns, local slope variations, and spectral curvature, respectively. The three representations are processed by separate branches and combined through bounded learnable residual fusion, with the raw spectrum serving as the primary representation and the derivatives providing complementary corrections. As a plug-and-play component, RDSF was integrated into Prototypical Network (ProtoNet), Relation Network (RelationNet), and Model-Agnostic Meta-Learning (MAML). The module was evaluated using spectra from 11,000 individual soybean seeds representing 11 varieties under known-class and strict class-disjoint unseen-class protocols. Under the representative known-class 3-way 10-shot setting with 15 query samples per class, RDSF increased the meta-test accuracy of RelationNet from 0.8898 ± 0.0201 to 0.9184 ± 0.0060. Under the unseen-class protocol, RDSF consistently improved ProtoNet and RelationNet across all evaluated shot settings; the largest gain was observed for ProtoNet in the 5-shot setting, with the meta-test accuracy increasing from 0.8848 ± 0.0253 to 0.9094 ± 0.0080. In contrast, RDSF did not consistently improve MAML under this protocol, indicating that its effectiveness depended partly on the underlying meta-learning mechanism. Ablation experiments and architecture comparisons further showed the complementary contributions of the derivative branches and the advantages of bounded residual fusion over a three-channel architecture and direct feature concatenation. Overall, RDSF provides an effective spectral representation module for metric-based few-shot classification of soybean seed varieties under the evaluated known-class and unseen-class conditions. Full article

To address the challenges of limited annotated datasets, strong demand for few-shot adaptation, and insufficient feature representation of traditional backbone networks in Mars rover onboard surface scene classification, this study proposes a classification framework that integrates a channel attention mechanism with meta-transfer learning. The work consisted of two main components. First, a dedicated annotated dataset for core Mars rover onboard surface scenes was constructed based on publicly available Mars Science Laboratory (MSL) Mastcam RGB imagery, providing a standardized experimental benchmark. Second, a classification model was developed based on ResNet50. Transfer learning with ImageNet pre-trained weights was employed to improve feature initialization, and an Squeeze-and-Excitation (SE) attention module was incorporated to enhance channel-wise feature representation. Furthermore, an appropriate activation function was selected under engineering constraints. Finally, a meta-learning strategy based on Model-Agnostic Meta-Learning (MAML) was introduced to improve adaptation capability in few-shot scenarios. Experimental results showed that the SE-ResNet model achieved a validation accuracy of 95.52%, representing an improvement of 17.41% over the baseline model. After integrating meta-transfer learning, the proposed SE-ResNet-MTL (SE-Enhanced ResNet Meta-Transfer Learning) model achieved accuracies of 90.5% and 91.5% in 1-shot and 5-shot tasks, respectively, outperforming traditional fine-tuning methods by 42.8% and 25.4%. These improvements were obtained with only a 5.47% decrease in full-dataset accuracy. Overall, the proposed method effectively balances full-dataset performance and few-shot generalization, providing a practical and efficient solution for Mars rover onboard surface scene recognition and other extraterrestrial visual tasks. Full article

Cutaneous metastases are an uncommon but clinically significant manifestation of internal malignancy, occurring in approximately 0.7–10.4% of patients with cancer. In some cases, they represent the first clinical manifestation of an otherwise occult malignancy, making accurate diagnosis critical for timely patient management. Distinguishing cutaneous metastases from primary cutaneous neoplasms, particularly malignant adnexal tumors, remains one of the most challenging problems in dermatopathology because of their substantial clinical, histopathologic, and immunophenotypic overlap. This review summarizes the epidemiology and biology of cutaneous metastasis and the clinical presentation of cutaneous metastases and emphasizes the central role of clinicopathologic correlation in diagnosis. Histologic features such as a purely dermal or subcutaneous location, intravascular tumor emboli, and a “bottom-heavy” growth pattern favor metastasis, whereas the presence of an in situ component or transition from a benign precursor lesion strongly supports a primary cutaneous neoplasm. We review the diagnostic utility of optimized immunohistochemical panels, highlighting the complementary roles of p63, cytokeratin 15, calretinin, and D2-40 (podoplanin) in establishing primary adnexal lineage, together with emerging markers including SOX10, androgen receptor, TRPS1, adipophilin, INSM1, SATB2, and BerEP4 for diagnostically challenging cases. We also discuss recent advances in molecular biology and comprehensive genomic profiling, including recurrent gene fusions (e.g., MYB::NFIB, CRTC1::MAML2, and YAP1 fusions) and characteristic mutational signatures that provide increasingly robust diagnostic evidence of tumor lineage. Finally, we provide a comprehensive, practical diagnostic algorithm for differentiating primary cutaneous adnexal carcinomas from cutaneous metastases of adenocarcinomas. By integrating traditional histopathologic techniques with modern immunohistochemical and molecular techniques, pathologists and clinicians can successfully navigate this complex differential diagnosis and thereby facilitate appropriate patient management and therapeutic intervention. Full article

Deep learning has exhibited potential in land use and land cover (LULC) classification applications. However, the effectiveness of deep learning remains constrained by the availability and quality of annotated training data. The persistent scarcity of labeled samples and spatial heterogeneity of remote sensing imagery hinder the robustness and generalization of trained models. Meta-learning, commonly referred to as “learning to learn”, is a paradigm that trains models over a distribution of tasks to acquire transferable knowledge, enabling rapid adaptation to new tasks with only a few labeled samples. This cross-task learning capability makes meta-learning a promising solution to data scarcity and spatial heterogeneity in the remote sensing context. This paper provides a systematic review of meta-learning applications in LULC classification, identifying a total of 70 relevant studies between 2018 and 2025. Three mainstream meta-learning paradigms (memory-augmented, optimization-based, and metric-based) are reviewed, and the applications are analyzed across four core challenges in LULC remote sensing: label scarcity, cross-region and cross-domain distribution shifts, temporal dynamics modeling, and multimodal data integration. The review reveals that optimization-based and metric-based methods dominate current research, with MAML and its variants being the most widely adopted due to the model-agnostic property, while memory-augmented methods remain underexplored. A consistent finding is that meta-learning outperforms conventional pre-training followed by fine-tuning under significant domain shifts across multiple data modalities. Current limitations, including computational overhead, episodic training constraints, and the lack of standardized evaluation protocols, are discussed. Future directions in cross-domain generalization, integration with foundation models, novel architectures, and standardized benchmarks are identified. Full article

Federated learning offers a promising approach for cross-institutional financial risk control modeling but encounters two key challenges in practice: feature space heterogeneity and low sample overlap rate. Current federated transfer learning methods often rely heavily on sufficient overlapping samples or explicit feature alignment. However, these approaches frequently result in negative transfer when enforced alignment is applied in highly heterogeneous environments. To address this issue, we propose FedVI, a novel federated transfer learning framework that integrates tabular-to-image conversion and meta-learning mechanisms. Moving beyond conventional methods that rely on sample-level alignment, FedVI employs a federated dual-stream feature alignment strategy to securely reconstruct a unified global feature map across institutions. Subsequently, FedVI integrates federated Image Generator for Tabular Data (IGTD) with tabular Transformer technology to convert one-dimensional tabular data into two-dimensional visual-semantic tensors. These tensors effectively fuse spatial topology and semantic information while embedding an independent Mask channel to explicitly retain the true missingness patterns of features. Finally, FedVI adopts the Model-Agnostic Meta-Learning (MAML) architecture to facilitate global parameter optimization. We evaluated FedVI on the real-world Lending Club credit dataset and Home Credit Default Risk datasets under highly heterogeneous federated settings (i.e., heterogeneous feature spaces across three clients and scarce overlapping samples). The results reveal that FedVI achieves competitive performance against advanced baselines such as FedProx, FedRep, and FedKT, particularly in recall and F1-Score. These findings indicate that FedVI can effectively support cross-domain adaptation under heterogeneous federated learning settings. Full article

Faced with the challenges of cross-cultural communication of public opinion in emergency events, traditional sentiment recognition methods struggle to accurately capture the complex semantics under multi-lingual and multi-symbol systems. This paper takes the powerful 7.7-magnitude earthquake that struck Myanmar in 2025 as a case study. It constructs a multi-dimensional public opinion annotation framework that integrates four types of semantic information—time, space, subject, and sentiment—by extracting data from multi-source textual materials, including social media, news reports, and government announcements. Building on this foundation, we design an improved Model-Agnostic Meta-Learning (MAML) model that incorporates cultural features to enhance sentiment recognition performance in low-resource cross-linguistic scenarios. Experimental results show that the model outperforms traditional methods in terms of sentiment classification accuracy, cultural semantic deviation rate and metaphor recognition ability. Furthermore, the research reveals the coupling mechanism of public opinion communication of “cultural modulation–agenda game”, and clarifies the influence paths and weight distributions among multiple subjects. The research results provide theoretical support and practical paths for improving the governance capacity of cross-border public opinion in emergency events and the construction of multilingual monitoring models. Full article

Deploying highly generalizable deepfake detection systems on resource-constrained edge devices poses a significant technical challenge for conventional end-to-end large models that rely heavily on computational resources. Extracting multi-source physical prior features is a viable approach under limited computational power; however, in few-shot scenarios, the dimensional mismatch of heterogeneous features is prone to causing downstream classifiers to overfit. To mitigate this bottleneck, this paper proposes a “static feature extraction–central normalization alignment–independent downstream decision” decoupled detection system for few-shot cross-domain tasks on edge devices. The front end of the system constructs an 856-dimensional comprehensive feature reservoir, and a lightweight residual normalization adapter $g_{\varphi }$ is introduced as the central support module. This module explicitly compresses the intra-class variance of heterogeneous features, providing a smoothly aligned manifold base for downstream classifiers. Experimental results indicate that this decoupled architecture demonstrates consistent stability in few-shot ($K=10$) cross-domain evaluations. When encountering intra-family cross-domain shifts and cross-mechanism distribution shifts from diffusion models, the accuracy reaches 84.9% and 76.1%, respectively. Compared to representative end-to-end meta-learning baselines (e.g., MAML), the relative error rate is reduced by over 30%. Furthermore, after completing the asynchronous offline pre-processing (approximately 897 ms) at the front end, a single-image online classification query requires only 7.7 ms under a simulated single-core CPU constraint, satisfying the low-latency requirements for lightweight deployment on edge devices. Finally, combined with empirical observations, this paper discusses the performance boundaries of the architecture in cross-mechanism metric mismatch scenarios, providing a low-barrier, robust engineering defense scheme for resource-constrained environments. Full article

Model-Agnostic Meta-Learning (MAML) and Prototypical Networks (ProtoNet) establish the foundational paradigms for few-shot classification. However, MAML suffers from optimization instability caused by reconstructing classification boundaries for every new task. Conversely, ProtoNet lacks the internal mathematical capacity necessary for task-specific parameter adaptation under domain shifts. To reconcile these structural limitations, we introduce Metric-based Model-Agnostic Meta-Learning (M²AML). By completely excising the parameterized classification layer from the episodic adaptation sequence, our framework replaces traditional inner-loop classification with a dynamic self-exclusive geometric similarity metric. Substituting functional mappings with spatial distance optimizations efficiently resolves evaluation conflicts, thereby establishing perfectly synchronized inner and outer learning rates alongside substantially accelerated adaptation steps. Extensive experiments across mini-ImageNet, tiered-ImageNet, and CIFAR-FS validate our approach against a comprehensive array of established algorithms. To ensure strictly fair comparative evaluations, we meticulously reproduce the MAML, ProtoNet, and Proto-MAML baselines. Empirical results demonstrate that M²AML achieves state-of-the-art performance across most evaluation settings, delivering absolute accuracy improvements ranging from 0.1% to 2.1% over existing leading models. Full article

Traditional malware classification approaches primarily address fixed sets of well-studied malware types and therefore struggle to accommodate the continual emergence of novel or previously unseen malware strains. While visualization-based strategies have shown promise in few-shot malware classification, existing methods often produce representations with limited semantic richness. In parallel, few-shot learning models frequently converge with suboptimal solutions, limiting their ability to generalize effectively to new classes. To address these challenges, we propose MetaWave, a unified framework that jointly optimizes both data representation and model learning for few-shot malware classification. Rather than treating feature representation and learning strategy as largely independent stages, MetaWave is formulated as an explicit representation–adaptation integration framework that combines multi-view malware encoding with meta-learning-based optimization. At the data level, we propose a Wavelet Transform-based Malware Representation method that leverages multi-scale frequency analysis and complementary views to generate semantically enriched representations. At the model level, we adopt Model-Agnostic Meta-Learning (MAML) to optimize model initialization for rapid adaptation to unseen tasks under limited data conditions. Extensive experiments are conducted on two benchmark datasets, EMBER and Malicia, under a 5-way 5-shot protocol with disjoint class splits to ensure evaluation on previously unseen malware families. The proposed framework achieves superior performance, reaching 97.8% accuracy on EMBER and 96.2% on Malicia, consistently outperforming state-of-the-art methods. These results indicate that jointly enhancing representation quality and model adaptability can improve classification accuracy and unseen-family performance under the evaluated 5-way 5-shot protocol. Overall, MetaWave provides an effective framework for few-shot malware classification and offers a promising basis for detecting emerging malware under limited-data conditions, while robustness to adversarial perturbation, obfuscation, and polymorphism remains to be validated through dedicated future evaluation. Full article

Motor imagery (MI) classification using EEG signals is central to brain–computer interfaces but remains challenging due to low signal-to-noise ratio, non-stationarity, and high inter-subject variability. We introduce EEG-TriNet++, a multi-branch deep learning architecture that enhances both classification accuracy and cross-subject generalization. The model integrates three complementary components: convolutional spatial–spectral encoders for channel-wise and frequency-specific patterns, bidirectional LSTMs to model temporal dynamics, and a Transformer head for global relational reasoning. A patchwise tokenization strategy and neural architecture search optimize the trade-off between efficiency and representational capacity. To address individual differences, a model-agnostic meta-learning (MAML) module enables rapid adaptation to new users with limited data. Evaluated on two public MI datasets under within-subject and leave-one-subject-out (LOSO) protocols, EEG-TriNet++ achieves 79.1% and 78.6% accuracy in within-subject tasks, and 72.4% and 71.3% in LOSO settings. Ablation studies validate the contribution of each module, and comparisons with state-of-the-art methods demonstrate consistent performance gains under identical conditions. Full article

As network communication technologies rapidly advance, ransomware has emerged as a significant cybersecurity threat that organizations cannot ignore. Static analysis enables rapid identification of ransomware by examining file structure and code characteristics before execution. However, existing classifiers are predominantly designed under the closed-set assumption, causing them to misclassify novel variants into known families. Furthermore, ransomware datasets typically exhibit long-tailed distributions with emerging families having very few available samples, making it difficult for models to learn discriminative features. To address these challenges, we propose Few-Shot Open-Set Ransomware Detection through Meta-learning and Energy-based Modeling (MEM), a unified open-set recognition framework based on static analysis of Portable Executable features. By integrating Model-agnostic Meta-learning (MAML), the model rapidly adapts to new families with limited samples. The Energy Function quantifies the confidence of predictions in distinguishing between known samples and unknown ones, while Focal Loss dynamically adjusts sample weights to reduce bias introduced by imbalanced distributions. The experimental results demonstrate that MEM achieves higher classification accuracy and better rejection performance of unknown samples than existing open-set recognition methods. Full article

Background/Objectives: The emergence of FLT3 inhibitors (FLT3i) has radically transformed the prognostic and therapeutic landscape for FLT3-mutated Acute Myeloid Leukemia, stimulating the need for comprehensive and structured clinical guidance. Methods: We aimed to develop evidence-based recommendations spanning the entire disease continuum of FLT3-mutated AML from leading Italian experts through a modified Delphi consensus process. Results: The panel achieved a high degree of agreement on specific interventions covering diagnostic testing, upfront FLT3i integration, role of allogeneic hematopoietic cell transplantation (allo-HSCT), Minimal Residual Disease (MRD) monitoring, and relapsed/refractory (R/R) strategies. Key recommendations mandate that analysis for both FLT3-ITD and FLT3-TKD mutations is required at diagnosis, with capillary electrophoresis or NGS as preferred methods. All fit patients with FLT3m-AML must receive intensive chemotherapy plus a FLT3i (midostaurin or quizartinib) and be evaluated for allo-HSCT. For unfit patients, the current standard of HMA + venetoclax is considered suboptimal, making the search for alternative strategies imperative. MRD monitoring using available molecular or flow cytometry markers is recommended to assess relapse risk and to optimize the allo-HSCT strategy. In the R/R setting, retesting the FLT3 status is mandatory, and gilteritinib is the standard treatment, serving as a bridge-to-transplant and for post-HSCT maintenance. Conclusions: The integration of FLT3i has shifted FLT3m-AML into a more favorable intermediate prognostic category, enhancing the role of curative strategies like allo-HSCT. This consensus paper provides a structured evidence-based comprehensive guide, translating complex data into clear actionable clinical recommendations that minimize practice variability and ultimately optimize management for this high-risk population. Full article

With the rapid development of genomic big data and genome-wide association study technologies, massive genomic data are available for the genetic dissection, development and utilization of important economic traits. Various GWAS algorithms have become increasingly efficient, enabling high-performance processing of these massive datasets. This has made it possible to conduct genetic dissection of economic traits based on big data and advanced statistical methods, which will provide accurate target loci for future trait improvement and genetic manipulation, greatly accelerating the process of genetic breeding. In this study, genotyping of 426 fish was performed using the T7 sequencing platform and 555,242 SNPs distributed across all the chromosomes were screened by data cleaning. We compared the performance of two GWAS methods, GCTA and GEMMA, in both single-trait and multi-trait frameworks. Twenty-nine SNPs significantly associated with seven traits were identified through single and multi-trait combined GWAS. Single-trait GWAS analysis using GCTA identified 1047 and 1452 significant loci for six growth traits and one sex trait (phenotypic sex, male or female) respectively, ultimately revealing 10 candidate genes, including slc48a1a, filip1L, nedd9, Crebbpa, LOC134024622, zbtb18, LOC117378376, LOC131530706, syde2, and col24a1. Similarly, 671 and 642 significant SNPs were detected with GEMMA for single-trait GWAS associated with six growth traits and the sex trait, respectively. In total, 16 candidate genes were mapped for these seven traits. Multi-trait GWAS was also performed using GEMMA for the six growth traits (sex was included as a covariate). The traits were grouped into five combinations based on their genetic correlations. A total of 37 SNPs were identified, corresponding to 10 candidate genes: LOC131530706, LOC134022516, abat, maml3, cica, LOC124013321, slc25a12, dnah10, syt9a, and LOC136932979. Notably, five overlapping candidate genes (LOC131530706, LOC134022516, abat, slc25a12 and dnah10) were also identified in both single- and multi-trait GWAS methods of GEMMA, highlighting their genetic stability and significance. The two GWAS methods, GCTA and GEMMA, identified two genes that were the same. The results of this study provide molecular markers and genetic resources for the improvement of growth traits in Plecoglossus altivelis. Full article

The rapid expansion of Internet of Things (IoT)-enabled smart grids has intensified the need for reliable fault detection and autonomous self-healing under non-stationary operating conditions characterized by frequent concept drift. To address the limitations of static and single-strategy adaptive models, this paper proposes H-CLAS, a novel Hybrid Continual Learning for Adaptive Self-healing framework that unifies regularization-based, memory-based, architectural, and meta-learning strategies within a single adaptive pipeline. The framework integrates convolutional neural networks (CNNs) for fault detection, graph neural networks for topology-aware fault localization, reinforcement learning for self-healing control, and a hybrid drift detection mechanism combining ADWIN and Page–Hinkley tests. Continual adaptation is achieved through the synergistic use of Elastic Weight Consolidation, memory-augmented replay, progressive neural network expansion, and Model-Agnostic Meta-Learning for rapid adaptation to emerging drifts. Extensive experiments conducted on the Smart City Air Quality and Network Intrusion Detection Dataset (NSL-KDD) demonstrate that H-CLAS achieves accuracy improvements of 12–15% over baseline methods, reduces false positives by over 50%, and enables 2–3× faster recovery after drift events. By enhancing resilience, reliability, and autonomy in critical IoT-driven infrastructures, the proposed framework contributes to improved grid stability, reduced downtime, and safer, more sustainable energy and urban monitoring systems, thereby providing significant societal and environmental benefits. Full article

Biphenotypic sinonasal sarcoma (BSNS) is a very rare, locally aggressive sarcoma arising in the sinonasal region, initially recognized as low-grade sinonasal sarcoma with neural and myogenic differentiation. Here, we report a case of BSNS extending into the intracranial and intraorbital regions, finally diagnosed by a break-apart fluorescence in situ hybridization (FISH) assay for rearrangements of PAX3. A 50-year-old woman presented with left diplopia and exophthalmos. CT and MRI revealed a large ethmoidal mass with intracranial and intraorbital extension. Since preoperative biopsy suggested a benign tumor, endoscopic endonasal resection was performed while preserving the anterior skull base and intraorbital structures. Postoperative histopathological diagnosis indicated synovial sarcoma, and proton beam therapy with adjuvant chemotherapy was subsequently administered. After treatment, FISH demonstrated rearrangements of PAX3 and MAML3 genes, leading to a revised diagnosis of BSNS, which typically does not require chemotherapy due to its non-metastatic behavior. Eleven years after treatment, the patient remains free of recurrence. Understanding BSNS is essential to avoid excessive intervention, and confirmation of PAX3 rearrangement by FISH or equivalent molecular testing is crucial for accurate diagnosis. Full article