Search Results (1,602)

Malicious packages in open source-software supply chains are a growing security concern, yet machine learning detectors built on registry metadata are difficult to interpret and are typically evaluated under protocols susceptible to data leakage. We construct a dataset of 3330 package versions from NPM and PyPI in which every attribute is reconstructed at its exact publication timestamp, and we partition the data so that all versions of a package, all packages of a maintainer, and all members of a name-similarity family reside within a single fold. The primary analysis is restricted to NPM, where 18 of 456 point-in-time attributes attain an AU-PR of 0.9972 against a positive-class baseline of 0.8755 and an AUC-ROC of 0.9828 on 763 held-out versions. A secondary pooled analysis over both ecosystems is reported: the class priors differ by nearly a factor of nine, a classifier using only the ecosystem of origin attains an AUC-ROC of 0.8036, and the pooled figures should be read with that composition in mind. Confidence intervals are obtained by cluster-bootstrap resampling at the composite-group level and are 3.7-times wider than version-level intervals in AU-PR (0.0201 against 0.0055); at that scale, none of the differences between feature subsets, selection-stage orderings, or metadata families reported here is distinguishable from sampling variation. At an operational prevalence of 0.1%, the positive predictive value is 4.3% and recall at the selected threshold is 0.537, indicating that metadata-based detection functions as a triage filter rather than a definitive verdict. Full article

BOOLE is a four-face educational platform integrating analog, combinational-logic, and sequential-logic activities with optional AI-assisted component identification and datasheet support. The system was developed through requirements translation, circuit simulation, two-layer PCB design, mechanical review, fabrication, assembly, functional verification, and iterative refinement. A Raspberry Pi 5, Camera Module 3 NoIR, and touchscreen support image capture and local interaction, while an Arduino Mega provides deterministic control of the physical learning faces. Segmented power energizes only the selected face and activity, and removable boards improve maintenance and fault isolation. Hardware demonstrations reproduced the intended voltage-divider, diode threshold/polarity, counter, and sequential-logic states. Ten one-versus-rest classifiers were fine-tuned from a pretrained ViT-Base model using 2000 original photographs, with 200 images for each of ten categories. The dataset was partitioned class-wise into mutually exclusive 80/10/10 training, validation, and final-test sets before augmentation, which was applied only to training data. Final-test accuracy ranged from 91.0% to 99.5%, with precision, recall, F1-score, specificity, balanced accuracy, and confusion matrices also evaluated. A 73-student pilot produced 89–96% positive (Yes) responses across six binary survey items, providing preliminary evidence of learner-perceived effectiveness, engagement, usability, and theory-to-practice support. Overall, BOOLE demonstrates a feasible, serviceable architecture for progressive electronics education. Full article

At the Lower Cretaceous Phra Prong locality in eastern Thailand, isolated spinosaurid teeth show two distinct morphotypes: one with well-marked apicobasal enamel ribs and one without ribs, with the enamel showing only a subtle wrinkling. The latter morphotype is unusual, since most spinosaurid teeth from Asia show a well-marked ribbing. The occurrence of two tooth morphotypes at Phra Prong is interpreted as corresponding to the presence of two distinct spinosaurid taxa. These two morphotypes probably indicate different feeding habits and niche partitioning. However, the functional significance of enamel ribbing (or its absence) remains poorly understood. Full article

Ecological evidence for epifaunal communities on offshore wind farm structures remains geographically concentrated in temperate Northeast Atlantic systems, whereas comparable studies from the subtropical Northwest Pacific remain limited. We sampled upper, middle, and lower layers of five pile-supported structures in a shallow offshore wind farm in the northern South China Sea during spring and autumn, integrating taxon-specific wet biomass, occurrence data, environmental variables, beta-diversity partitioning, and eight functional traits. Taxon richness, functional trait-combination richness, and total wet biomass were lower in the upper layer than in the middle and lower layers, while wet-biomass-weighted Shannon diversity was lower in the upper than in the lower layer. Season and layer × season effects were not significant. Beta-diversity partitioning showed contributions of both turnover and nestedness-resultant dissimilarity, with relative contributions varying descriptively among seasons and layer contrasts. Measured environmental variables explained limited community variation, and the overall db-RDA was not significant, whereas variation partitioning identified the vertical layer as the only predictor group with a significant independent contribution. Global RLQ and FDR-corrected fourth-corner analyses did not support trait–environment coupling. These results extend evidence of vertical epifaunal differentiation to a geographically underrepresented subtropical Northwest Pacific offshore wind farm and support multi-layer sampling for biofouling assessment. Full article

Artificial intelligence can extend the functional capabilities of embedded sensors by extracting application-level information from physical measurements. This study investigates whether footstep-induced floor vibrations acquired with a single triaxial MEMS accelerometer contain sufficient information to characterize pedestrian path orientation and travel sense. A custom sensing platform based on an ADXL355 accelerometer and an ESP32 microcontroller was developed to acquire the structural vibration response at 4 kSPS. Lightweight temporal features were processed using a Random Forest classifier. The primary assessment used leakage-aware event-level cross-validation, with complete footsteps as the data-partitioning units. Under this more conservative protocol, discrimination of the complete A–K movement-label set was poor, whereas a compact 12-dimensional descriptor representation achieved 73.18% accuracy, 71.52% balanced accuracy, and 70.98% macro-F1 for X/Y path-orientation classification. Reliable positive/negative travel-sense discrimination could not be demonstrated from isolated footsteps. For historical comparison, the original sample-level procedure yielded 97.09% accuracy, but this value is retained only as a within-sequence reference because densely sampled observations contain strongly overlapping information. The findings provide proof-of-concept evidence that a single floor-mounted MEMS accelerometer can capture coarse pedestrian path-orientation information without cameras or spatially distributed vibration-sensor networks. Feature extraction and classifier inference were performed offline; broader validation across participants, sessions, floor structures, and realistic disturbances is required before deployment as an embedded edge AI sensing node. Full article

The cold supply chain (CSC) reverse logistics segment remains underexplored in the risk assessment literature. Existing frameworks apply traditional failure mode and effects analysis (FMEA) with a three-criterion structure (severity, occurrence, detectability), which inadequately captures the time-sensitive and cost-intensive risk environment of CSC reverse logistics. This study proposes an integrated framework combining hazard analysis and critical control points (HACCP)-based node partitioning, an extended five-criterion FMEA incorporating timeliness and economic cost, triangular interval-valued fuzzy number (TIVFN) aggregation via the Aczél–Alsina operator, and cumulative prospect theory (CPT) to prioritize risk modes. Applied to a food-sector reverse CSC, 16 failure modes were identified across six HACCP-based process nodes. Severity and economic cost jointly account for 47.5% of total criterion weight. Monitoring and data management (node P6) consistently emerge as the dominant risk cluster, with data completeness (FM14), sensor accuracy (FM15), and early-warning functionality (FM16) occupying the top three positions across all weight scenarios. Relative to conventional models and the extended RPN, the TIVFN-CPT model assigns FM16 a substantially higher priority, demonstrating that CPT captures asymmetric, loss-averse risk perception that conventional methods fail to encode. Sensitivity analysis across five weight scenarios confirms the structural robustness of the rankings. By extending FMEA to a five-criterion behavioral framework and introducing TIVFNs as an uncertainty-preserving linguistic scale, this study indicates that information quality constitutes a risk factor of comparable priority to physical temperature maintenance, supporting a tiered resource allocation strategy that prioritizes investment at node P6 and targeted upgrades at nodes P1 and P4. Full article

Metabolic dysfunction-associated steatotic liver disease (MASLD) is the most common chronic liver disease worldwide, encompassing a spectrum from simple steatosis to metabolic dysfunction-associated steatohepatitis (MASH), fibrosis, cirrhosis and hepatocellular carcinoma. Increasing evidence indicates that disease progression is driven not by hepatic triglyceride accumulation alone but by the metabolic partitioning of excess fatty acids between adaptive and maladaptive pathways. Ketogenesis, traditionally viewed as a fasting-induced mechanism for disposing of excess acetyl-CoA, is now recognized as a key regulator of hepatic metabolic homeostasis, coordinating mitochondrial substrate utilization, carbon flux and systemic metabolic adaptation. In addition to serving as oxidative fuels, ketone bodies, particularly β-hydroxybutyrate, function as signalling metabolites that modulate inflammation, oxidative stress, mitochondrial function and epigenetic regulation. Despite increased fatty acid delivery in obesity and insulin resistance, ketogenic capacity becomes progressively impaired during MASLD, promoting mitochondrial acetyl-CoA accumulation, oxidative stress and diversion of carbon toward lipotoxic lipid synthesis while reducing protective β-hydroxybutyrate signalling. This review examines ketogenesis as an integrative metabolic checkpoint linking fatty acid oxidation, lipid metabolism, mitochondrial function and immune signalling in MASLD. We discuss how impaired ketogenic flux contributes to hepatocellular injury, fibrosis and metabolic inflexibility, and evaluate the therapeutic potential of restoring ketogenesis to prevent disease progression. Full article

Reliable short-term position forecasting may provide anticipatory state information for collision-risk assessment, communication management, and prediction-assisted control in Internet of Things (IoT)-enabled unmanned aerial vehicles (UAVs); these downstream functions are not evaluated directly here. This study reformulates UAV position prediction as a flight-wise, multi-horizon, three-dimensional forecasting problem and tests whether position, velocity, gravity-resolved acceleration, and quaternion-orientation histories improve predictive accuracy while measuring model-level edge-inference cost rather than end-to-end system latency. The dataset contains 3100 AirSim flights with high-rate kinematic, inertial, attitude, pressure, and magnetic-field measurements under variable horizontal wind. The reported generalization is flight-disjoint within one AirSim domain; route/scenario disjointness and transfer to physical UAVs are not established. Signals are converted to a common navigation frame, gravity-resolved, low-pass filtered, resampled to 50 Hz, and partitioned by flight identifier before normalization and window construction. Each learned model receives 2 s of history and predicts the complete next 1 s trajectory, with errors evaluated at 0.1, 0.5, and 1.0 s. The sensor-enriched LSTM (LSTM-PVAQ) is compared under matched conditions with persistence, constant-velocity, constant-acceleration, extended Kalman filter, reduced-feature LSTM, GRU, temporal convolutional network (TCN), and compact Transformer baselines. LSTM-PVAQ achieved 3D RMSE values of 0.043, 0.168, and 0.371 m at 0.1, 0.5, and 1.0 s, respectively. At 1 s, its RMSE was 21.7% lower than LSTM-PV, 13.1% lower than GRU-PVAQ, 9.3% lower than TCN-PVAQ, and 16.8% lower than Transformer-PVAQ. Its one-second ADE and FDE were 0.216 and 0.339 m. On a Raspberry Pi 5 CPU using one FP32 thread and batch size one, median neural forward-pass latency was 0.88 ms, well below the 20 ms model-update interval. The results show that gravity-resolved inertial and orientation histories improve multi-horizon prediction, while TCN-PVAQ remains an attractive lower-latency alternative. Full article

This paper proposes a novel mixed numerical scheme for approximating linear Fredholm integral equation systems (LFISs) using a combination of Bernoulli polynomials (BPs) and enhanced block-pulse functions $\left(\mathcal{E}\mathcal{B}\mathcal{P}Fs\right)$. This suggested representation makes use of both the local support nature and computation efficiency of the $\left(\mathcal{E}\mathcal{B}\mathcal{P}Fs\right)$ as well as the high-order approximating nature of BPs. Using the operational matrices, the system of coupled integrals can be transformed into a finite-dimensional algebraic system (AS) of expansion coefficients. A theoretical analysis is established to investigate the solvability, convergence, stability, and approximation error of the resulting scheme. In addition, the effects of polynomial degree and partition refinement on the numerical accuracy are examined. Several test problems are considered, and the obtained results demonstrate that the proposed $\mathfrak{B}\mathcal{E}\mathcal{B}\mathcal{P}\mathcal{F}$ approach provides highly accurate approximations while requiring relatively small basis dimensions. Comparisons with previously reported numerical techniques further illustrate their computational effectiveness and accuracy. Full article

Graph labeling is an important area of graph theory that studies the assignment of integers to the vertices or edges of a graph according to specific rules. Among these labeling methods, distance magic labeling has attracted considerable attention due to its interesting combinatorial structure and applications in network design and communication systems. A distance magic labeling of a graph is a bijection from the vertex set to the set $\{1,2,\dots ,n\}$ such that the sum of the labels of the neighbors of each vertex is equal to a constant called the magic constant. This paper investigates the existence and construction of distance magic labelings for certain families of complete bipartite graphs. Two principal cases are studied, namely graphs of the form $K_{2m,2m}$ and $K_{2m-1,2m}$. For graphs of the form $K_{2m,2m}$, an explicit construction is developed showing that these graphs admit a distance magic labeling for every integer $m\ge 1$. The corresponding magic constant is derived as $k=m(4m+1)$ and the validity of the construction is verified by proving bijectivity of the labeling function and equality of vertex weights. A similar constructive approach is applied to graphs of the form $K_{2m-1,2m}$, where distance magic labelings are obtained using structured arithmetic label distributions. These constructions are further extended by applying vertex swapping techniques and block-based arguments to generate complete bipartite graphs $K_{p,q}$ that preserve the same magic constant for certain values of p and q. These findings contribute to the understanding of how arithmetic structure and partition properties influence the existence of distance magic labelings and suggest several directions for further research in graph labeling theory. Full article

Salt stress restricts cucumber seedling establishment by impairing root development, photosynthesis, ion homeostasis, and redox balance. This study examined whether potassium silicate (K₂SiO₃) could partially alleviate these responses under hydroponic sodium chloride (NaCl) stress. Cucumber seedlings were exposed to six treatments: a nutrient-solution control, K₂SiO₃ alone supplying 1.0 mmol L⁻¹ silicon (Si), 75 mmol L⁻¹ NaCl, and NaCl combined with K₂SiO₃ supplying 0.5, 1.0, or 2.0 mmol L⁻¹ Si. Growth, root morphology, photosynthetic pigments, gas exchange, chlorophyll fluorescence, photosystem II (PSII) energy partitioning, oxidative injury, antioxidant enzyme activities, osmotic adjustment, and ion status were evaluated at 7 and 14 d. NaCl markedly reduced seedling growth, root development, net photosynthetic rate, PSII photochemical performance, and electron transport, while increasing leaf sodium (Na⁺), malondialdehyde accumulation, non-photochemical quenching, non-regulated energy loss, proline, and soluble sugar. K₂SiO₃ partially alleviated these changes. The treatment supplying 1.0 mmol L⁻¹ Si produced the strongest integrated recovery of growth, root activity, photosynthetic performance, PSII function, and oxidative status. The treatment supplying 2.0 mmol L⁻¹ Si resulted in the lowest leaf Na⁺ concentration and the highest leaf potassium (K⁺)/Na⁺ ratio among salt-stressed seedlings but did not produce the greatest growth recovery. These findings suggest coordinated changes in photosynthesis, photochemical energy use, redox status, and ion balance. Because K₂SiO₃ supplied both Si and K⁺, the results represent responses to K₂SiO₃ supplementation rather than Si-specific effects. Full article

Transformer lifecycle data across organizations are typically vertically partitioned among material suppliers, manufacturers, logistics service providers, testing agencies, and operation and maintenance units. This study presents a reproducible multi-layer vertical federated learning (VFL) prototype that integrates salted-hash identifier matching, additive sharing of local score vectors over finite fields, and a local public key infrastructure with a signature-based audit verification mechanism. A deterministic synthetic dataset is first constructed, comprising 5200 aligned records and 31 predictor variables, which are partitioned among five participants with varying numbers of features per participant. Second, across five validation runs, the VFL models under both the standard block-wise and score-sharing paths achieved an AUC of 0.8825 ± 0.0119, an F1 score of 0.7367 ± 0.0249, and an accuracy of 0.8102 ± 0.0183 on the test set. The classification results of both paths were fully consistent with the centralized gradient-descent logistic regression baseline. Notably, the score-sharing path exhibited a maximum log-odds deviation of only 2.22 × 10⁻⁸ on the test set, with no prediction discrepancies observed. Third, across 10 independently generated synthetic populations, the nonlinear output mechanism highlights the limitations of linear models: the AUC of vertical federated learning (VFL) drops to 0.6457 ± 0.0171, while Extra Trees and HistGradientBoosting achieve 0.7731 ± 0.0149 and 0.7743 ± 0.0139, respectively. Finally, in a separate residual-sharing diagnostic test, when 1 to 4 participants jointly shared the residuals, the label inference AUC remained around 0.499–0.500; however, when all five participants shared or plaintext residuals were used, the labels could be fully recovered. Both simple membership inference diagnostic tests yielded results close to random. The local signature log verifier rejected all 700 injected faults and accepted the 400 clean control log events. These results validate the feasibility of numerical reproducibility and local audit functionality under synthetic data and single-process conditions, yet they are insufficient to demonstrate end-to-end label privacy protection, malicious security, effectiveness on real data, or real-time ledger performance. Full article

Sustainable nutrient management in sweet potato requires matching fertilizer inputs with site-specific soil nutrient supply while maintaining productivity and processing quality. This study characterized site-specific NPK responses and identified fertilization regimes coordinating yield and starch content under contrasting soil nutrient backgrounds. To address this objective, a three-factor, five-level quadratic orthogonal rotatable composite design comprising 23 N–P–K experimental runs was conducted independently at two sites in Beibei and Youyang, Chongqing, China, using the starch-type sweet potato cultivar ‘Yushu 17’. The two sites differed markedly in initial soil nutrient status. Photosynthetic characteristics, nutrient accumulation, dry matter production and partitioning, fresh storage-root yield, and quality were measured. Quadratic regression models combined with a desirability function were used to characterize site-specific nutrient responses and optimize yield and starch content. The effects of N, P, and K differed markedly between sites. In Beibei, fresh storage-root yield was mainly affected by the linear effect of N, while P and K also had positive effects, and the P × K interaction was significant. In Youyang, yield was primarily regulated by the linear effect of K and showed a significant negative quadratic response to P. Photosynthetic performance; N, P, and K accumulation; total dry matter accumulation (TDMA); and root-to-top ratio (R/T) responded mainly to N in Beibei but were more sensitive to K in Youyang. Fresh storage-root yield was positively correlated with net photosynthetic rate, stomatal conductance, transpiration rate, chlorophyll content, leaf area index, nutrient accumulation, R/T, and TDMA, but negatively correlated with starch content, indicating a yield–starch trade-off. When yield was prioritized while starch content was maintained, the optimal N–P₂O₅–K₂O rates were 155.02–116.42–300.00 kg·ha⁻¹ in Beibei, predicting 42,800 kg·ha⁻¹ yield and 20.09% starch, and 153.34–75.00–282.90 kg·ha⁻¹ in Youyang, predicting 35,850 kg·ha⁻¹ yield and 24.92% starch. These site-specific optima provide a quantitative basis for more targeted fertilizer allocation while coordinating yield and starch content, thereby supporting sustainable nutrient management of starch-type sweet potato. Full article

We study prediction intervals for the semi-functional partial linear model (SFPLM) under stationary, geometrically $\beta$-mixing dependence. We analyze a split conformal procedure based on a three way data partition with buffer gaps, a functional principal component projection semi-metric on the functional covariate, profiled least squares estimation of the parametric component, kernel estimation of the nonparametric component and of the conditional standard deviation, and a studentized absolute residual score. Marginal validity of split conformal prediction with a trained score under $\beta$-mixing is available from generic results of Oliveira et al. and of Barber and Pananjady, without any buffer and without accuracy requirements on the fitted estimators. Our main result is complementary to those guarantees: a finite sample marginal lower coverage bound whose theoretical finite-sample coverage penalty decomposes additively into seven interpretable components expressed in the structural primitives of the SFPLM, quantifying the price of replacing the ideal SFPLM score by the estimated score inside the proof. The penalty plays no role in the computation of the interval, involves unknown structural constants, and is not an operational correction. The bound requires no parametric error model, but it is not assumption free; it holds under explicit structural conditions, including geometric $\beta$-mixing, conditionally centered sub-Gaussian errors, a fractal small ball regime for the projected functional covariate, and local regularity of the score distribution. Simulations under a protocol frozen before outcome computation, spanning mild and strong dependence, a misspecification stress test, and a dependent-score design with exactly quantified score autocorrelation, show near nominal coverage for all methods, with the gapped and contiguous variants statistically indistinguishable in coverage. Studentization showed no systematic coverage advantage, while interval-length differences were systematic. The value of the analysis lies in the explicit model-specific estimation layer of the coverage decomposition, not in a numerical gain over naive split conformal. Full article

The Min pig, a Chinese indigenous breed, is valued for its excellent meat quality and stress tolerance. Nevertheless, its growth rate falls considerably short of commercial pig breeds, a pattern typical of most unselected indigenous populations, and marked individual variation further undermines its economic viability. To explore the genetic basis of this variation, we evaluated a series of nonlinear fixed-effects and mixed-effects models based on the Gompertz, Logistic, and von Bertalanffy functions using body weight records from 92 Min pigs. Model selection was based on AIC, BIC, and leave-one-out cross-validation. The best-fitting model was a Logistic nonlinear mixed-effects model with individual-level random effects on all three growth parameters (Asymptotic weight, timing parameter, and be parameter), which clearly outperformed models with simpler random-effect structures and fixed-effects models. From this model, we identified three characteristic growth transition points: the early transition point (Growth Rate Index, GRI) at 94.83 days (22.32 kg), the single inflection point (Maximum Growth Rate, MGR) at 166.06 days (52.80 kg), and the late transition point (Late Growth Rate Index, LGRI) at 237.29 days (83.29 kg), with a maximum absolute growth rate of 488.2 g/day. These points partitioned growth into initial acceleration, rapid growth, deceleration, and Asymptotic growth phases. Using individual fitted growth curves, we selected five fast-growing and five slow-growing pigs that reached approximately 90 kg during the plateau phase, defined as a predicted body weight of at least 95% of the individual Asymptotic weight. The fast-growing group reached 90 kg at 240.5 ± 20.42 days, whereas the slow-growing group reached the same weight at 286.67 ± 19.20 days. At the 90 kg slaughter weight, we collected longissimus dorsi muscle samples from these pigs during the plateau phase and performed RNA-seq. Transcriptome analysis revealed 864 differentially expressed genes between the two groups, with 540 upregulated and 324 downregulated in the fast-growing group. Pathway enrichment implicated the PI3K-Akt and TGF-β signaling pathways in muscle development, and differential expression of IGFN1, DCN, COL3A1, MYOC, COL1A2, COL1A1, and IGF2 may explain the growth variation between the groups. In summary, the Logistic mixed-effects model with individual-level random effects on all growth parameters effectively captures the growth pattern of Min pigs, and the PI3K-Akt and TGF-β pathways likely mediate growth differences in this breed. Full article