Search Results (313)

The outcomes of a standard geochemical, geophysical and petrographical approach to exploration at Lead Trial, a small prospect in central Scotland, have been compared to the interpretation of a parallel gold compositional study describing 703 gold particles from local in situ and alluvial occurrences. Standard exploration approaches identified a 4.5 km² zone hosting an array of numerous auriferous (to 17 g/t Au), vuggy, brecciated quartz-galena ± sphalerite veins culminating in the identification of a drill target. The gold study identified three gold compositional types: two 23–32 wt.% Ag alloys with a Zn-Pb-Cu mineral inclusion assemblage differentiated by sphalerite abundance, and a 5–16 wt.% Ag alloy with a Mo-Bi-Pb-Cu-Fe inclusion signature, yet to be correlated with either float or outcrop. Spatial distribution of the gold types indicates lateral variation and probably vertical variation within a single magmatic hydrothermal system. Integration of gold particle studies with early stages of exploration offers rapid insights into the nature and distribution of mineralization when very limited information is available and is mutually supportive of standard exploration approaches. Full article

Accurate reconstruction of complete tree point clouds is essential for estimating ecosystem structural characteristics from LiDAR data. In urban forestry environments, however, terrestrial laser scanning (TLS) and mobile laser scanning (MLS) frequently produce incomplete observations. Occlusion caused by neighboring trees, together with interference from surrounding urban objects such as buildings and vehicles, often leads to missing regions within scanned point clouds. These defects may further affect the reliability of tree structural analysis and parameter estimation. Although recent learning-based point cloud completion methods have improved reconstruction performance, several limitations remain when they are applied to complex tree structures. Many existing networks depend on farthest point sampling (FPS) for feature extraction, which can result in the loss of fine-scale branching information. Furthermore, local feature aggregation methods based on the traditional k-nearest neighbor (KNN) strategy are highly sensitive to regions with uneven point cloud distribution, such as the canopy region where density variations are significant in tree point clouds. To alleviate these issues, this study proposes GS-TreeAttn, an attention-guided framework specifically for tree point cloud completion. This network models density and structural representation as a coupled problem and employs a structure-guided density-adaptive attention mechanism to jointly capture global structural dependencies and local geometric features. We comprehensively evaluate the proposed method using publicly available datasets and urban forestry data collected under real-world scanning conditions. Experimental results show that even in complex scenarios with severe occlusion and uneven sampling density, GS-TreeAttn generates more complete reconstruction results. This improvement is particularly evident in regions where the canopy and branches mutually occlude each other, where information loss is very common in real-world urban forestry. Full article

Feature selection represents a critical step in developing accurate and interpretable models for early breast cancer detection. Despite extensive research in the field of mammographic image analysis, no consensus has yet been reached on the optimal feature subsets that distinguish normal from malignant tissues. To address this gap, the present study aims to identify the most discriminative and significant features through a comprehensive multi-criterion selection framework. The aim is to integrate, as new frameworks, different combinations of t-test, ANOVA, Mutual Information (MI), and Equal Grouping Methods (EGM) to rank 19 linear and nonlinear features extracted from mammographic images. The objective is to maximize feature relevance while minimizing redundancy and enhancing diagnostic and healthcare systems. Linear features were assessed alongside nonlinear descriptors. A framework combining t-test, ANOVA, and EGM, guided by MI relevance, was employed to balance feature contributions across categories. The experimental results demonstrated that hybrid feature selection significantly enhanced diagnostic accuracy using optimal linear and nonlinear attributes. The optimization results suggested using a hybrid of six linear and eight nonlinear features. Linear features were highly accurate for detecting cancer. Haralick entropy obtained the highest average accuracy and performance, 94.14% and 93.45%; followed by kurtosis, 93.49% and 92.59%; perimeter irregularity, 93.43% and 92.65%; skewness, 93.01% and 92.25%; and volume/area, 92.82% and 91.92%. Despite the reliable discriminative power of linear descriptors, their overall effectiveness in representing intricate tissue characteristics was limited. The comparison of statistical characteristics shows a distinct performance benefit of nonlinear descriptors over linear ones for detecting breast cancer. Nonlinear descriptors, however, showcased higher accuracy and performance, with an average accuracy of 97.81% in contrast to 94.43% for linear approaches. Local phase congruency achieved the top average accuracy and performance, 97.81% and 96.61%, respectively; succeeded by wavelet entropy, 97.62% and 96.42%; Laplacian spectrum features, 97.52% and 96.32%; nonlinear diffusion, 97.10% and 95.90%; and clustering coefficient, 96.70% and 95.50%; then Shannon, Tsallis, and Rényi entropies. The results indicate that statistically validated nonlinear characteristics significantly outperform linear ones across accuracy and performance measures. Their combination significantly improves the strength and discriminative power of computer-assisted breast cancer diagnostic systems, affirming their suitability for integration into sophisticated machine learning and deep learning models. The results also show that the new multi-criterion framework’s early detection performance surpassed that of the statistical and deep learning models explored, with an average of 98.6% accuracy, 98% sensitivity, 98.9% precision, and 98.4% F1 score of early detection of breast cancer. The incorporation of statistically validated nonlinear descriptors, particularly local phase congruency and wavelet entropy, improves the discriminative ability, robustness, and clinical understanding of breast cancer computer-assisted diagnostic systems. Overall, the proposed framework confirms that integrating hybrid features substantially enhances robustness and plays a pivotal role in computer-assisted breast cancer detection. These selected features may be fed to more advanced algorithms in the future, potentially yielding improved performance. Full article

Federated learning lets organizations train a shared model without pooling private data. The standard method, Federated Averaging, requires all participants to use the same input features, a condition that fails in cross-sector phishing detection, where banks analyze URL structure and hospitals analyze email content. We present RankBridge, a system that groups participants by comparing ranked lists of SHapley Additive exPlanations (SHAP) feature importance rather than model weights or gradients. Each participant trains a local LightGBM model, extracts the top-K features by SHAP importance, and sends a 60-byte ranked list of feature indices to a central server. The server applies rank correlation and Ward’s hierarchical clustering to identify similarly threatened organizations. RankBridge operates in two modes: ModelShare, where models are also shared within each discovered group for prediction ensembling, and RankOnly, where the server returns only a group label and each participant keeps their model private. Across 32 participants in five organization types, RankBridge (ModelShare) achieves F1 $=0.853$ (AUC $=0.926$) on synthetic data and F1 $=0.772$ (AUC $=0.812$) on real phishing data, and it is the only method to outperform isolated local training on both. On real heterogeneous data the standard baselines adapted to LightGBM, including Federated Averaging, retain a moderate thresholded F1 (≈0.73) but their ranking quality collapses to near-random (AUC $\approx 0.59$, PR-AUC $\approx 0.66$), whereas RankBridge sustains AUC $=0.812$ and PR-AUC $=0.819$. RankBridge recovers the correct organizational groupings with Normalized Mutual Information (NMI) $=0.973$. The rank-based grouping channel itself transmits 60 bytes per participant per round, roughly 10,000× less than a full model upload. Full article

RGB-D salient object detection (SOD) aims to segment the most prominent objects from the background with a pair of given RGB and depth images. Existing RGB-D methods usually rely on heavy backbones to achieve high accuracy, while current lightweight methods struggle to maintain competitive performance. To break this intractable trade-off between effectiveness and model complexity, we propose a Lightweight Spatial-Frequency Collaborative Interaction Network (SFCINet), a unified and highly efficient framework. The core of SFCINet resides in the synergy between spatial-domain features and frequency-domain global priors. Specifically, we introduce the Spatial-Frequency Synergy (SFS) module, which shifts the perspective to a joint complex Fourier domain. By adaptively learning and optimizing the decoupled amplitude and phase components, it effectively isolates clutter to yield a purified global frequency-synergized prior, which modulates the spatial branches to eliminate cross-modal discrepancies for subsequent feature fusion while supplementing global information during decoding. To alleviate the interference caused by cross-modal representation discrepancies, we design the Cross-Guidance Interaction (CMGI) module, which employs a reciprocal anchoring mechanism. It guides the counterpart to mutually filter irrelevant noise and select task-relevant information, achieving fusion in an efficient manner. Finally, we present a Calibrated Hierarchical Decoder (CHD), which injects frequency-synergized global priors into the hierarchical decoding process. It re-establishes the connection between the frequency and spatial domains, ultimately achieving global-local consistency. Extensive experiments demonstrate that SFCINet delivers superior performance over state-of-the-art methods. Full article

Hyperspectral image (HSI) classification enables precise classification of land-cover types from rich spectral and spatial information. Recent methods combine convolutional neural network (CNN) and Mamba branches to exploit their complementary local and global modeling capabilities for HSI classification. However, most of these methods treat all spectral channels uniformly in feature fusion, failing to account for the discriminability differences across spectral bands. Moreover, most methods rely on a single classification head at the final layer, which may lead to vanishing gradients in shallow layers. To address these limitations, a spectral group-wise gated CNN–Mamba network with cross-stage mutual distillation, called SGGCMNet, is proposed. To address the first limitation, a CNN–Mamba spectral group-wise gating block (CMSB) is designed at the feature-fusion level. Specifically, the CMSB partitions channels into multiple sub-groups along the spectral dimension. Each sub-group learns its own fusion weights that balance local spectral–spatial cues produced by a CNN pathway with long-range context produced by a Mamba pathway. To address the second limitation, two loss-level optimization strategies are proposed jointly: A progressive deep supervision strategy with uncertainty-based dynamic weighting is proposed to attach classification heads at all network stages. A temperature-regulated cross-stage mutual-distillation mechanism is further designed to enable bidirectional knowledge transfer among classification heads at different stages. On three benchmark HSI datasets, SGGCMNet achieves state-of-the-art accuracy. Full article

Reliable landslide mapping in active mining regions is constrained by two coupled issues: severe domain shift from public datasets and extremely limited local annotations. In line with Transformer-centric intelligent interpretation of complex remote-sensing scenes, this study proposes a label-efficient transfer segmentation framework from a public source corpus to target open-pit mines built on SegFormer with a lightweight hybrid adapter that couples global context modeling with mining-specific directional cues. The pipeline combines source-domain Transformer pre-training, class-conditional feature alignment, Bayesian uncertainty estimation, and human-guided active learning. First, the backbone is pre-trained on the GDCLD source domain to learn transferable landslide morphology priors. Second, a joint optimization stage with class-conditional alignment reduces source and target embedding discrepancy during adaptation. Third, Monte Carlo dropout is enabled at inference to estimate predictive distributions, and sample acquisition is driven by mutual-information-based querying to prioritize epistemically informative target patches, addressing the small-sample supervision challenge emphasized in remote-sensing deep learning. This design turns uncertainty into an operational annotation policy rather than a passive diagnostic output. Experimental results show that the framework consistently outperforms deterministic counterparts and strong active-learning baselines in spectrally complex mine scenes, while approaching the fully supervised upper bound with only a small fraction of local labels. The approach is especially effective in shadowed benches and fault-adjacent slopes, supporting trustworthy deployment for geohazard monitoring and disaster-relevant slope safety workflows; extension to multi-modal constraints (e.g., SAR or elevation) is discussed as future work. Full article

Multimodal omics data portray biological processes across molecular layers, yet their heterogeneity and high dimensionality hinder a unified representation. Existing integrative approaches either focus on local feature interactions or adopt static fusion, often overlooking the complementary global sequential context and the dynamic relevance among omics sources. Consequently, clinically critical tasks such as accurate cancer-subtype classification and therapy selection still lack sufficient accuracy and robustness. We introduce the Dual-Branch Contrastive Learning for Multi-Omics Dynamic Fusion Network (DBCL-DFNet), a dual-branch contrastive-learning framework that simultaneously encodes local heterogeneous graphs and global omics sequences, distills key features via contrastive objectives, and employs a dynamic attention mechanism for adaptive, data-driven fusion. Benchmarked on three public cancer multi-omics datasets, DBCL-DFNet outperforms both conventional machine-learning models and state-of-the-art deep-integration methods, establishing a competitive and reliable framework for multi-omics integration and demonstrating potential for precision-oncology decision-making. From an information-theoretic perspective, the framework integrates Copula-entropy-guided feature selection with mutual-information-maximizing contrastive alignment, providing a principled foundation for robust multi-omics integration. Full article

Alzheimer’s disease (AD) is a multifactorial neurodegenerative disorder characterized by the simultaneous disruption of interconnected molecular pathways, yet the structural mechanisms underlying this transcriptional disintegration remain poorly characterized. To address this, we constructed condition-specific gene co-expression networks from DLPFC bulk RNA-seq data, using a mutual-information (MI) framework with infomap community partitioning. Functional enrichment of network communities via Ingenuity Pathway Analysis (IPA) identified GABAergic signaling, SNARE complex assembly, Synaptogenesis, and neurexin and neuroligin interactions as significantly overrepresented pathways. Integration of node degree with condition-specific average expression revealed coordinated topological centralization of key synaptic genes—including NRXN2, LRRTM1, DLGAP3, and SHANK1—alongside a widespread transcriptional downregulation in GABAergic and Synaptogenesis modules. A shortest-path analysis revealed a consistent expansion of intra-pathway distances across all evaluated canonical pathways in AD, a pattern statistically consistent with reduced local co-expression cohesion. These findings reframe Late-Onset Alzheimer’s Disease (LOAD) as an active structural-rewiring process, in which the observed topological centralization pattern seems to be consistent with a consolidation of co-expression around synaptic components, though we cannot exclude that shifts in cellular composition contribute to this signal in bulk RNA-seq data. Full article

Accurate identification of Chinese sensitive words is critical for maintaining online information security. However, this task faces three technical challenges: (1) high contextual dependency causing semantic ambiguity; (2) adversarial variations (e.g., homophones, character splitting) that evade exact matching; and (3) scarcity of high-quality annotated samples in complex scenarios, leading to few-shot distribution characteristics. To address these challenges, we propose a Branch Transformer Mutual-Information Calibration (BTMC) network. Specifically: (i) to capture multi-level, cross-dimensional semantic interactions despite limited data, we design a branch-based Transformer structure that aligns and fuses features across different semantic dimensions; (ii) to establish context channels between global and local semantics under few-shot conditions, we introduce a global-local interactive fusion mechanism that enhances focus on core semantics; (iii) to improve discriminability of complex semantic patterns, we propose a semantic calibration regularization mechanism that reweights features and balances information distribution. Experimental results on a newly constructed Chinese sensitive words dataset (45,623 sentences, four categories) demonstrate that BTMC achieves average F1-scores of 0.9715 (Politics and Violence), 0.9683 (Rudeness and Vulgarity), 0.9704 (Drugs and Gambling), and 0.9531 (Others), outperforming state-of-the-art baselines by 10–15% relative improvement. The code and dataset will be made publicly available. Full article

Storage location assignment in automated warehouses strongly affects order-processing efficiency. Existing co-occurrence-based approaches often rely on pointwise mutual information (PMI) statistics or direct frequency co-occurrence. This paper compares two deliberately chosen representation families for storage assignment in automated warehouses operated under Multi-Agent Pickup and Delivery (MAPD) constraints: Pointwise Positive PMI (PPPMI), representing direct pairwise co-occurrence, and Latent Dirichlet Allocation (LDA), representing latent-topic smoothing. The purpose is not to benchmark every possible representation space, but to make the pairwise-versus-latent contrast interpretable under a fixed execution pipeline consisting of task construction, visit-order selection, path planning, and collision avoidance. The broader research setting is motivated by real warehouse order data in which SKU co-occurrence structure is present, but such logs mix latent-topic effects, explicit family-based co-occurrence, noise, and demand variation. We therefore use two controlled abstractions of order structure: one generator with latent-topic mixtures and one generator with more direct family co-occurrence. We embed the proposed LDA representation and the PPPMI baseline in constrained-clustering and simulated-annealing placement methods and evaluate them against frequency-based, load-balancing, and random baselines. Evaluation is conducted in a fixed extended MAPD simulator that explicitly models orientation-aware motion, turning costs, service times, dynamic task release, and collision avoidance. In the latent-topic regime, LDA-based methods tended to form the leading group in average finite-horizon makespan, computed over completed combinations of random seeds and operating conditions. In the supplementary direct-co-occurrence condition, PPPMI was competitive in the plain representation comparison, while LDA-driven local search on top of a frequency-based initial layout remained strong. These results do not imply that LDA is universally superior; rather, they indicate that the relative suitability of PPPMI and LDA depends on the order structure and on how the representation interacts with the placement optimizer. The controlled generators are useful for isolating those effects, but they do not replace external validation on real warehouse logs. Full article

This article offers an exploratory overview of the contemporary Ethiopian textile, fashion and apparel system. The contribution originated from a teaching experience in Addis Ababa within the framework of the AICS–UNIDO-funded project “Ethiopia: Support to Youth and Women through Products and Services Development and Public–Private Partnerships in the Fashion Value Chain” which prompted the authors to deepen their understanding of the local fashion ecosystem. Drawing on informal conversations, observations, and ethnographically oriented field notes, the authors developed the analysis through desk research and a review of the relevant literature. The picture that emerges reveals both the creativity and strong entrepreneurial drive of Ethiopian designers, alongside the structural barriers they commonly face, including limited access to materials, investment, and institutional support. Designers are shown to negotiate ongoing tensions between cultural heritage and global aesthetics, while also contending with local consumption patterns situated between second-hand clothing markets and international brands. These dynamics highlight both the challenges and the potential of the Ethiopian fashion scene, pointing to opportunities for mutual learning and for fostering fashion practices that are sustainable, globally relevant, and firmly grounded in local contexts. Full article

While Graph Autoencoders (GAEs) have become a standard for unsupervised representation learning, their reliance on integer-order convolutions inherently restricts information propagation to immediate local neighborhoods. This paper introduces the Fractional Graph Autoencoder (FGAE) and its variational extension (FVGAE) to move beyond these local constraints. By integrating fractional Laplace operators, our framework generalizes conventional GAEs and enables tunable non-local propagation. We show that the fractional order $\alpha$ acts as a structural regularizer, utilizing the Green’s function of anomalous diffusion to induce a form of structural memory within the latent space. This allows the model to recover long-range dependencies that are typically lost in standard architectures. Systematic benchmarking across eight datasets—ranging from homophilic citation networks to heterophilic and dense product graphs—shows that these fractional variants consistently outperform both foundational and state-of-the-art baselines (ARGA, SIG-VAE, and GraphMAE). Notably, on the Amazon Computers and Citeseer datasets, our methods achieve relative increases in Normalized Mutual Information (NMI) of 77.55% and 67.28%, respectively. Statistical analysis confirms these gains are robust, with large effect sizes (Cohen’s $d>0.80$) and significance at $p<0.05$. These findings suggest that fractional graph autoencoding offers a mathematically grounded inductive bias for capturing the complex, multi-scale dynamics of real-world networked systems. Full article

As multi-view datasets expand across diverse practical fields, feature selection (FS) has become an indispensable preparatory stage for machine learning models. Nevertheless, real-world multi-view data is often unlabeled and distributed among isolated clients, posing significant challenges to traditional centralized methods due to privacy concerns and communication constraints. Furthermore, existing centralized and federated approaches frequently suffer from entrapment in local optima and lack robust convergence guarantees. To address these issues, we propose Fed-MUFSHT, a federated framework for multi-view unsupervised FS (MUFS) that integrates tensor learning with a novel metaheuristic optimizer, Hierarchical-Cognitive Tianji’s Horse Racing Optimization (HC-THRO). Within the federated learning paradigm, Fed-MUFSHT follows a dual-stage local optimization process. Stage 1 applies HC-THRO, which integrates Hierarchical Competitive Learning and Adaptive Cognitive Mapping to simulate multi-level strategic competition and cognitive adaptation among individuals. This design enhances global exploration, adaptive learning, and fine-grained feature selection in high-dimensional spaces. Stage 2 employs a TL module based on canonical polyadic (CP) decomposition to perform missing-view imputation and refine latent representation learning. At the global level, a privacy-preserving aggregation strategy based on Normalized Mutual Information (NMI) and feature weights enables efficient model coordination without exposing raw data. Comparative experiments on several public benchmark datasets reveal that Fed-MUFSHT maintains clear advantages over strong competing methods, showing better optimization results together with more dependable convergence characteristics. The overall evidence suggests that the proposed approach is both robust and effective for distributed optimization tasks involving privacy protection. Full article

Corner detection is a fundamental task in computer vision that plays a critical role in applications such as image registration, 3D reconstruction, and object tracking. In biological visual systems, simple cells in the primary visual cortex exhibit high selectivity to edge stimuli of specific orientations, while end-stopped cells can detect geometric singular structures such as line segment endpoints and corners. Existing corner detection methods based on visual neural computation typically employ a strategy of densely distributed end-stopped cells for corner localization, which suffers from significant localization deviation under small angle conditions due to mutual interference between responses of adjacent neurons. To address this problem, this paper proposes a bionic corner detection method based on cooperative processing of simple cells and end-stopped cells. The method constructs a two-stage cooperative processing framework: the edge filtering stage employs a Gabor filter bank to simulate the orientation selectivity of simple cells, extracting edge positions and orientation information; the dynamic construction stage builds unilateral end-stopped cells only at filtered edge positions based on local orientation information, fundamentally avoiding computational redundancy and response interference caused by global dense distribution; the corner localization stage determines precise corner coordinates through hierarchical clustering and dual-cluster centroid fusion strategies. Experimental results demonstrate that, in the 15° acute-angle regime where dense end-stopped schemes are most severely affected by response interference, the proposed method reduces the mean localization error from 8.76 to 2.34 pixels, corresponding to a 73.3% improvement; averaged across the eight tested angle levels from 15° to 165°, the improvement is approximately 40.9%, and all per-angle differences are statistically significant (paired t-test, p < 0.01 or below, N = 10 independent runs). On standard test images, the method attains the lowest mean localization error among the eight compared detectors (1.58 pixels, versus 1.68–3.42 pixels for Harris, FAST, COSFIRE, KAZE, SuperPoint, Deep Corner, and Wei et al.), while maintaining competitive detection rate, false-alarm rate, and runtime. Physiological plausibility validation experiments show that the correlation coefficient between the detection deviation of this method and human perceptual deviation reaches 0.923, indicating that the output of the framework aligns with previously reported human perceptual bias patterns and supporting its biological plausibility as a biologically inspired—rather than mechanistic—model of corner perception. The source code, dataset, and experimental results are publicly available (see Data Availability Statement). Full article