Search Results (126)

Protein embedding is a protein representation that carries along the information derived from filtering large volumes of sequences stored in large archives. Routinely, the protein is represented by a matrix in which each residue is a context-specific vector whose dimensions reflect the size of the large architectures of neural networks (transformers) trained with deep learning algorithms on large volumes of sequences. A recently introduced method (Embedding-Based Alignment, EBA) is particularly suited for pairwise embedding comparisons and, as we report here, allows for remote homolog detection under specific constraints, including protein sequence length similarity. Multifunctional proteins are present in different species. However, particularly in humans, the problem of their structural and functional annotation is urgent since, according to recent statistics, they comprise up to 50% of the human reference proteome. In this paper we show that when EBA is applied to a set of randomly selected multifunctional human proteins, it retrieves, after a clustering procedure and rigorous validation on the reference Swiss-Prot database, proteins that are remote homologs to each other and carry similar structural and functional features as the query protein. Full article

Few-shot object detection in remote sensing imagery faces significant challenges, including limited labeled samples, complex scene backgrounds, and subtle inter-class differences. To tackle these issues, we design a novel detection framework that effectively transfers supervision from a few annotated support examples to the query domain. We introduce a feature enhancement mechanism that injects fine-grained support cues into the query representation, helping the model focus on relevant regions and suppress background noise. This allows the model to generate more accurate proposals and perform robust classification, especially for visually confusing or small objects. Additionally, our method enhances feature interaction between support and query images through a nonlinear combination strategy, which captures both semantic similarity and discriminative differences. The proposed framework is fully end-to-end and jointly optimizes the feature fusion and detection processes. Experiments on three challenging benchmarks, NWPU VHR-10, iSAID and DIOR, demonstrate that our method consistently achieves state-of-the-art results under different few-shot settings and category splits. Compared with other advanced methods, it yields superior performance, highlighting its strong generalization ability in low-data remote sensing scenarios. Full article

Temporal reasoning is an important part of the field of time geography and spatio-temporal data science. Recent advances in qualitative temporal reasoning have developed a set of 74 relations that apply between discretized time intervals of at least two pixels each. While the identification of specific relations is important, the field of qualitative spatial and temporal reasoning relies on conceptual neighborhood graphs to address relational similarity. This similarity is paramount for generating essential decision support structures, notably reasonable aggregations of concepts into single terms and the determination of nearest neighbor queries. In this paper, conceptual neighborhood graphs of qualitative topological changes, with discretized temporal interval relations in the form of translation, isotropic scaling, and anisotropic scaling, are identified using data generated through a simulation protocol. The outputs of this protocol are compared to the extant literature regarding conceptual neighborhood graphs of the Allen interval algebra, demonstrating the theoretical accuracy of the work. This work supports the development of robust spatio-temporal artificial intelligence as well as the future development of spatio-temporal query systems upon the spatio-temporal stack data architecture. Full article

Building high-quality multimodal instruction datasets is often time-consuming and costly. Recent studies have shown that a small amount of carefully selected high-quality data can be more effective for improving LVLM performance than large volumes of low-quality data. Based on these observations, we propose an error-guided multimodal sample selection framework with hallucination suppression for LVLM fine-tuning. First, semantic embeddings of queries are clustered to form balanced subsets that preserve task diversity. A visual contrastive decoding module is then used to reduce hallucinations and expose genuinely difficult examples. For closed-ended tasks, such as object detection, we estimate sample value using prediction accuracy; for open-ended question answering, we use the perplexity of generated responses as a difficulty signal. Within each cluster, high-error or high-perplexity samples are preferentially selected to construct a compact yet informative training set. Experiments on the InsPLAD detection benchmark and the PowerQA visual question answering dataset show that our method consistently outperforms random sampling under the same data budget, achieving higher F1, cosine similarity, BLEU (Bilingual Evaluation Understudy), and GPT-4o-based evaluation scores. This demonstrates that hallucination-aware, uncertainty-driven data selection can improve LVLM robustness and data efficiency. Full article

Generative artificial intelligence (GenAI) is rapidly permeating research practices, yet knowledge about its use and topical profile remains fragmented across tools and disciplines. In this study, we present a cross-disciplinary map of GenAI research based on the Web of Science Core Collection (as of 4 November 2025) for the ten tool lines with the largest number of publications. We employed a transparent query protocol in the Title (TI) and Topic (TS) fields, using Boolean and proximity operators together with brand-specific exclusion lists. Thematic similarity was estimated with the Jaccard index for the Top–50, Top–100, and Top–200 sets. In parallel, we computed volume and citation metrics using Python and reconstructed a country-level co-authorship network. The corpus comprises 14,418 deduplicated publications. A strong concentration is evident around ChatGPT, which accounts for approximately 80.6% of the total. The year 2025 shows a marked increase in output across all lines. The Jaccard matrices reveal two stable clusters: general-purpose tools (ChatGPT, Gemini, Claude, Copilot) and open-source/developer-led lines (LLaMA, Mistral, Qwen, DeepSeek). Perplexity serves as a bridge between the clusters, while Grok remains the most distinct. The co-authorship network exhibits a dual-core structure anchored in the United States and China. The study contributes to bibliometric research on GenAI by presenting a perspective that combines publication dynamics, citation structures, thematic profiles, and similarity matrices based on the Jaccard algorithm for different tool lines. In practice, it proposes a comparative framework that can help researchers and institutions match GenAI tools to disciplinary contexts and develop transparent, repeatable assessments of their use in scientific activities. Full article

Medical image segmentation, driven by the intrinsic fractal characteristics of biological patterns, plays a crucial role in medical image analysis. Recently, universal image segmentation, which aims to build models that generalize robustly to unseen anatomical structures and imaging modalities, has emerged as a promising research direction. To achieve this, previous solutions typically follow the in-context learning (ICL) framework, leveraging segmentation priors from a few labeled in-context references to improve prediction performance on out-of-distribution samples. However, these ICL-based methods often overlook the quality of the in-context set and struggle with capturing intricate anatomical details, thus limiting their segmentation accuracy. To address these issues, we propose VG-SAM, which employs a multi-scale in-context retrieval phase and a visual in-context guided segmentation phase. Specifically, inspired by the hierarchical and self-similar properties in fractal structures, we introduce a multi-level feature similarity strategy to select in-context samples that closely match the query image, thereby ensuring the quality of the in-context samples. In the segmentation phase, we propose to generate multi-granularity visual prompts based on the high-quality priors from the selected in-context set. Following this, these visual prompts, along with the semantic guidance signal derived from the in-context set, are seamlessly integrated into an adaptive fusion module, which effectively guides the Segment Anything Model (SAM) with powerful segmentation capabilities to achieve accurate predictions on out-of-distribution query images. Extensive experiments across multiple datasets demonstrate the effectiveness and superiority of our VG-SAM over the state-of-the-art (SOTA) methods. Notably, under the challenging one-shot reference setting, our VG-SAM surpasses SOTA methods by an average of $6.61\%$ in DSC across all datasets. Full article

The Retrieval-Augmented Generation (RAG) framework shows great potential in terms of improving the reasoning and knowledge utilization capabilities of language models. However, most existing RAG systems heavily rely on large language models (LLMs) and suffer severe performance degradation when using small language models (SLMs), which limits their efficiency and deployment in resource-constrained environments. To address this challenge, we propose Retrieval-Adaptive-Planning RAG (RAP-RAG), a lightweight and high-efficiency RAG framework with adaptive retrieval task planning that is compatible with both SLMs and LLMs simultaneously. RAP-RAG is built on three key components: (1) a heterogeneous weighted graph index that integrates semantic similarity and structural connectivity; (2) a set of retrieval methods that balance efficiency and reasoning power; and (3) an adaptive planner that dynamically selects appropriate strategies based on query features. Experiments on the LiHua-World, MultiHop-RAG, and Hybrid-SQuAD datasets show that RAP-RAG consistently outperforms representative baseline models such as GraphRAG, LightRAG, and MiniRAG. Compared to lightweight baselines, RAP-RAG achieves 3–5% accuracy improvement while maintaining high efficiency and maintains comparable efficiency in both small and large model settings. In addition, our proposed framework reduces storage size by 15% compared to mainstream frameworks. Component analysis further confirms the necessity of weighted graphs and adaptive programming for robust retrieval under multi-hop reasoning and heterogeneous query conditions. These results demonstrate that RAP-RAG is a practical and efficient framework for retrieval-enhanced generation, suitable for large-scale and resource-constrained scenarios. Full article

Background/Objectives: Solitary plasmacytoma of bone (SPB) results from abnormal proliferation of plasma cells and accounts for 2–5% of all plasmacytic malignancies. Radiation therapy is the standard of care in treating SPB due to its efficacy in controlling disease progression and optimizing patient survival. However, prior studies have highlighted disparities in radiation therapy receipt among various cancer types. In this study, we aim to investigate whether similar sociodemographic and clinical disparities exist in the treatment of SPB through use of the Surveillance, Epidemiology, and End Results (SEER) database. Methods: The SEER database was queried for biopsy-confirmed cases of SPB between 2000 and 2021 using the ICD-O-3 histology code 9731/3 and primary site codes C40.0–41.9. Chi-square tests, Fisher’s exact tests, and multivariable logistic regression were completed using SPSS v29.0.2, with significance set to p < 0.05. Results: A total of 4139 patients were identified, of which 75.3% received treatment with radiation therapy. Multivariable analysis revealed that low-income patients making less than $74,999 annually (aOR 0.80, 95% CI 0.67–0.97), as well as those from non-Hispanic Asian/Pacific Islander (aOR 0.49, 95% CI 0.33–0.73) and Hispanic (aOR 0.77, 95% CI 0.60–0.98) racial and ethnic groups, were significantly less likely to receive radiation therapy. Conclusions: These findings reveal notable disparities in radiation therapy utilization for SPB patients based on income and race and ethnicity, emphasizing the need for interventions to address systemic inequities, improve access to care, and ensure that all patients receive high-quality cancer care to optimize long-term outcomes. Full article

The purpose of few-shot relation extraction (RE) is to recognize the relationship between specific entity pairs in text when there are a limited number of labeled samples. A few-shot RE method based on a prototype network, which constructs relation prototypes by relying on the support set to assign labels to query samples, inherently leverages the symmetry between support and query processing. Although these methods have achieved remarkable results, they still face challenges such as the misjudging of noisy samples or outliers, as well as distinguishing semantic similarity relations. To address the aforementioned challenges, we propose a novel semantic enhanced prototype network, which can integrate the semantic information of relations more effectively to promote more expressive representations of instances and relation prototypes, so as to improve the performance of the few-shot RE. Firstly, we design a prompt encoder to uniformly process different prompt templates for instance and relation information, and then utilize the powerful semantic understanding and generation capabilities of large language models (LLMs) to obtain precise semantic representations of instances, their prototypes, and conceptual prototypes. Secondly, graph attention learning techniques are introduced to effectively extract specific-relation features between conceptual prototypes and isomorphic instances while maintaining structural symmetry. Meanwhile, a prototype-level contrastive learning strategy with bidirectional feature symmetry is proposed to predict query instances by integrating the interpretable features of conceptual prototypes and the intra-class shared features captured by instance prototypes. In addition, a clustering loss function was designed to guide the model to learn a discriminative metric space with improved relational symmetry, effectively improving the accuracy of the model’s relationship recognition. Finally, the experimental results on the FewRel1.0 and FewRel2.0 datasets show that the proposed approach delivers improved performance compared to existing advanced models in the task of few-shot RE. Full article

In recent years, indoor user identification via Wi-Fi signals has emerged as a vibrant research area in smart homes and the Internet of Things, thanks to its privacy preservation, immunity to lighting conditions, and ease of large-scale deployment. Conventional deep-learning classifiers, however, suffer from poor generalization and demand extensive pre-collected data for every new scenario. To overcome these limitations, we introduce SimID, a few-shot Wi-Fi user recognition framework based on identity-similarity learning rather than conventional classification. SimID embeds user-specific signal features into a high-dimensional space, encouraging samples from the same individual to exhibit greater pairwise similarity. Once trained, new users can be recognized simply by comparing their Wi-Fi signal “query” against a small set of stored templates—potentially as few as a single sample—without any additional retraining. This design not only supports few-shot identification of unseen users but also adapts seamlessly to novel movement patterns in unfamiliar environments. On the large-scale XRF55 dataset, SimID achieves average accuracies of 97.53%, 93.37%, 92.38%, and 92.10% in cross-action, cross-person, cross-action-and-person, and cross-person-and-scene few-shot scenarios, respectively. These results demonstrate SimID’s promise for robust, data-efficient indoor identity recognition in smart homes, healthcare, security, and beyond. Full article

We propose a method to address the challenge of course discovery on search platforms by employing large language models (LLMs) to parse extended search parameters from natural language queries. We developed a set of algorithms that augment a course search platform prototype by integrating an LLM-based assistant to facilitate 55,000 vocational training sessions. The developed method supports natural language queries and parses optional search parameters. For parameter optionality and to evaluate the feasibility of parameter parsing, we introduce a relevance check mechanism based on cosine similarity. The parsing process was conducted by using a guided generation strategy with grammar-based restrictions to limit the generation possibilities. The developed method enhanced the precision and pertinence of course searches. Full article

Vehicle re-identification (Re-ID) is a critical computer vision task that aims to match the same vehicle across spatially distributed cameras, especially in the context of remote sensing imagery. While prior research has primarily focused on Re-ID using remote sensing images captured from similar, typically elevated viewpoints, these settings do not fully reflect complex aerial-ground collaborative remote sensing scenarios. In this work, we introduce a novel and challenging task: aerial-ground cross-view vehicle Re-ID, which involves retrieving vehicles in ground-view image galleries using query images captured from aerial (top-down) perspectives. This task is increasingly relevant due to the integration of drone-based surveillance and ground-level monitoring in multi-source remote sensing systems, yet it poses substantial challenges due to significant appearance variations between aerial and ground views. To support this task, we present AGID (Aerial-Ground Vehicle Re-Identification), the first benchmark dataset specifically designed for aerial-ground cross-view vehicle Re-ID. AGID comprises 20,785 remote sensing images of 834 vehicle identities, collected using drones and fixed ground cameras. We further propose a novel method, Enhanced Self-Correlation Feature Computation (ESFC), which enhances spatial relationships between semantically similar regions and incorporates shape information to improve feature discrimination. Extensive experiments on the AGID dataset and three widely used vehicle Re-ID benchmarks validate the effectiveness of our method, which achieves a Rank-1 accuracy of 69.0% on AGID, surpassing state-of-the-art approaches by 2.1%. Full article

Generating adversarial examples under black-box settings poses significant challenges due to the inaccessibility of internal model information. This complexity is further exacerbated when attempting to achieve a balance between the attack success rate and perceptual quality. In this paper, we propose QTRL, a query-efficient two-phase reinforcement learning framework for generating high-quality black-box adversarial examples. Unlike existing approaches that treat adversarial generation as a single-step optimization problem, QTRL introduces a progressive two-phase learning strategy. The initial phase focuses on training the agent to develop effective adversarial strategies, while the second phase refines the perturbations to improve visual quality without sacrificing attack performance. To compensate for the unavailability of gradient information inherent in black-box settings, QTRL designs distinct reward functions for the two phases: the first prioritizes attack success, whereas the second incorporates perceptual similarity metrics to guide refinement. Furthermore, a hard sample mining mechanism is introduced to revisit previously failed attacks, significantly enhancing the robustness and generalization capabilities of the learned policy. Experimental results on the MNIST and CIFAR-10 datasets demonstrate that QTRL achieves attack success rates comparable to those of state-of-the-art methods while substantially reducing query overhead, offering a practical and extensible solution for adversarial research in black-box scenarios. Full article

Career planning agencies and other organizations can help workers if they are able to effectively identify related occupations that are relevant to the task at hand. Occupational knowledge bases such as O*NET and ESCO represent mature attempts to categorize occupations and describe them in detail so that they can be used to search for related occupations. Vector databases offer an opportunity to find related occupations based on large pre-trained word and sentence embeddings and their associated retrieval algorithms for similarity search. This paper reports a systematic empirical evaluation of the possibilities of using vector databases for related occupation retrieval using different document structures, embeddings, and retrieval configurations for two popular open source vector databases, and using the O*NET curated database. The objective was to understand the extent to which curated relations capture all the meaningful relations in a context of retrieval. The results show that, independent of the database used, distance metrics, sentence embeddings, and the selection of text fragments are all significant in the overall retrieval performance when comparing with curated relations, but they also retrieve other relevant occupations based on text similarity. Further, the precision is high for smaller cutoffs in the results list, which is especially important for settings in which vector database retrieval is set up as part of a Retrieval Augmented Generation (RAG) pattern. The inspection of highly ranked retrieved related occupations not explicit in the curated database reveals that text similarity captures the taxonomical grouping of some occupations in some cases, but also other cross-cuts different aspects that are distinct from the hierarchical organization of the database in most of the cases. This suggests that text retrieval should be combined with querying explicit relations in practical applications. Full article

Accurate segmentation of brain tumors in magnetic resonance imaging (MRI) remains a challenging task due to heterogeneous tumor structures, varying intensities across modalities, and limited annotated data. Deep learning has significantly advanced segmentation accuracy; however, it often suffers from sensitivity to hyperparameter settings and limited generalization. To overcome these challenges, bio-inspired metaheuristic algorithms have been increasingly employed to optimize various stages of the deep learning pipeline—including hyperparameter tuning, preprocessing, architectural design, and attention modulation. This review systematically examines developments from 2015 to 2025, focusing on the integration of nature-inspired optimization methods such as Particle Swarm Optimization (PSO), Genetic Algorithm (GA), Differential Evolution (DE), Grey Wolf Optimizer (GWO), Whale Optimization Algorithm (WOA), and novel hybrids including CJHBA and BioSwarmNet into deep learning-based brain tumor segmentation frameworks. A structured multi-query search strategy was executed using Publish or Perish across Google Scholar and Scopus databases. Following PRISMA guidelines, 3895 records were screened through automated filtering and manual eligibility checks, yielding a curated set of 106 primary studies. Through bibliometric mapping, methodological synthesis, and performance analysis, we highlight trends in algorithm usage, application domains (e.g., preprocessing, architecture search), and segmentation outcomes measured by metrics such as Dice Similarity Coefficient (DSC), Jaccard Index (JI), Hausdorff Distance (HD), and ASSD. Our findings demonstrate that bio-inspired optimization significantly enhances segmentation accuracy and robustness, particularly in multimodal settings involving FLAIR and T1CE modalities. The review concludes by identifying emerging research directions in hybrid optimization, real-time clinical applicability, and explainable AI, providing a roadmap for future exploration in this interdisciplinary domain. Full article