Search Results (452)

Transformer layers can develop distinct attention routing patterns during training, but how architectural access to structural information shapes this organization remains unclear. Existing structure-aware methods make this difficult to determine by modifying several components. To investigate this question, we compare two Transformers matched in field ID inputs, embeddings, learned components, backbone, parameter count, and optimization but differing in attention computation. Both use field embeddings; Structure-Aware Attention Bias (SAAB) additionally transforms the supplied field IDs into a fixed pairwise same-field attention bias. We first train both models on the DBpedia dataset with masked structure modeling (MSM), which requires the recovery of hidden field labels from the context. Both models exceed 99.9% validation accuracy, yet they organize attention differently. Across 64 paired validation examples from the seed-1001 model pair, SAAB changes the mean same-field attention mass (SFM) by $-0.085$ at Layer 2 and $+0.048$ at Layer 3 (95% confidence intervals: [$-0.092$,$-0.078$] and [$+0.042$,$+0.055$]). We call this pattern competitive displacement. Across eight initializations, two show relatively strong displacement; the others show weaker or different responses. For seed 1001, displacement varies non-monotonically with the bias strength. Related changes persist with depth in two selected seeds and appear differently on PubMed. These results show that SAAB redistributes layer-wise attention routing under tested conditions. Full article

Bluetooth Low Energy (BLE) resolvable private addresses (RPAs) reduce passive tracking, but RPA-resolving creates receiver-side work: a scanner whose bonded identity set exceeds controller resolving-list capacity which can be forced into repeated host-side Identity Resolving Key (IRK) searches by floods of syntactically valid unknown private-address candidates. We formulate this as capacity-constrained resolving scheduling and propose P3, our persistence-aware admission-control scheduler for BLE RPA resolution, evaluated in its main form P3-Persist. After cache and resolving-list fast paths miss and the unresolved-work budget is exhausted, P3 grants a small reserve only to over-the-air RPA values that reappear. This uses visible repetition rather than an identity-correlated pre-resolution label. In paired simulations, identical visible traces produced identical initial pre-resolution decisions, but later modeled delay/defer features were distinguishable in some tested conditions; accordingly, no end-to-end side-channel-free claim is made. In the evaluated 20-seed, 1800 s simulation matrix (N = 512, RL = 8), P3-Persist keeps modeled unique-flood work in the same bounded approximately 45k AES-equivalent/min regime as budget-only limiting while increasing modeled legitimate resolution to 0.9569 under medium flood and 0.9557 under heavy flood, compared to 0.57–0.62 for budget-only baselines. A 20-seed epoch–loss–density experiment shows that service recovery is conditional rather than universal, and a denser adaptive-address search identifies an overlapping near-cap region around m = 320 and r = 1, with a highest confirmed mean of 146,196.29 AES-equivalent/min (96.69% of the modeled cap). These results establish bounded-work and conditional service-recovery behavior within the simulator; they do not establish production-firmware effectiveness, end-to-end side-channel freedom, on-device timing, energy, current, latency, or queueing behavior. Full article

Fluopyram is a broad-spectrum fungicide and nematicide used in foliar sprays and seed treatments. Despite its low volatility and moderate solubility, it is frequently detected beyond application sites in air, soil, and water. As a per- and polyfluoroalkyl substance (PFAS)-class “forever chemical”, its off-site presence raises ecotoxicological concerns. We conducted a 61-day greenhouse experiment to assess the effects of fluopyram residues in soil (0, 0.02, 0.04, and 0.78 mg kg⁻¹ soil) on two cover crop species (Italian clover, Trifolium incarnatum; cornflower, Centaurea cyanus), earthworms (Lumbricus terrestris), springtails (Folsomia candida), and litter decomposition. Measured endpoints included plant establishment, growth, biomass, photosynthetic efficiency, root architecture, soil fauna activity, and litter breakdown. Responses were species-specific and did not exhibit clear dose–response relationships. Height growth in Italian clover and cornflower showed significant treatment × date interactions, whereas establishment and biomass production remained unaffected. In Italian clover, root diameter increased with fluopyram concentration, while root length and volume showed non-significant trends. Cornflower roots exhibited fewer crossings at higher concentrations but were otherwise unaffected. Photosynthetic efficiency remained unchanged in both species, although it tended to decrease modestly in cornflower at the highest concentration. Springtail surface activity density was generally influenced by fluopyram, with significant treatment × date interactions at 0.04 and 0.78 mg kg⁻¹. Earthworm effects were inconclusive due to low survival rates. Neither litter decomposition rate nor stabilization factor was affected. Our results show that fluopyram residues in soil can induce subtle, species-specific effects in non-target organisms. However, long-term studies are needed to clarify how this highly persistent active ingredient and its degradation products accumulate in soils and interact with co-occurring pesticides, thereby improving ecological risk assessments. Full article

Background: Rheumatoid arthritis (RA) is a chronic autoimmune disease frequently accompanied by depression, with persistent inflammation, oxidative stress, and neuroimmune dysfunction contributing to disease progression. This study evaluated the therapeutic efficacy of a Gum Arabic-stabilized bioactive oil blend nanoemulsion and investigated its underlying mechanisms using molecular docking and network pharmacology. Methods: A lyophilized nanoemulsion prepared from grape seed oil, wheat germ oil, and avocado peel oil was characterized for physicochemical properties and phytochemical composition. Female rats with Freund’s complete adjuvant-induced rheumatoid arthritis received the nanoemulsion at two doses. Paw inflammation, behavioral performance, inflammatory cytokines, oxidative stress biomarkers, acetylcholinesterase concentration, lipid profile, and liver and kidney function were evaluated. Molecular docking and network pharmacology analyses were performed to identify potential molecular targets and signaling pathways. Results: The nanoemulsion exhibited favorable physicochemical characteristics and was rich in phenolic compounds, flavonoids, phytosterols, tocopherols, and unsaturated fatty acids. Treatment significantly reduced paw inflammation, TNF-α, IL-6, malondialdehyde, and acetylcholinesterase while enhancing catalase activity, improving metabolic parameters, and alleviating depression-like behavior. Molecular docking demonstrated favorable binding of the major phytochemicals to TNF-α, IL-6, and acetylcholinesterase. Network pharmacology identified key therapeutic targets, including TNF, AKT1, PTGS2, IL6, STAT3, ESR1, and NR3C1, and revealed enrichment of inflammatory, oxidative stress, and neuroimmune signaling pathways. Conclusions: The Gum Arabic-stabilized bioactive oil blend nanoemulsion ameliorated rheumatoid arthritis-associated depression-like behavior through a multi-component–multi-target–multi-pathway mechanism involving coordinated regulation of inflammatory, oxidative stress, cholinergic, and neuroimmune pathways. These findings support its potential as a complementary nutraceutical strategy for rheumatoid arthritis and its associated neurobehavioral complications. Full article

Security operations can detect that an intrusion is under way, yet they cannot say how long an ongoing attack still needs to reach a critical objective such as data exfiltration. Prior work on multi-stage attacks identifies the active stage or predicts the next step, but does not estimate the residual time to compromise from real traffic using survival models. This paper addresses that gap through a temporal framework built on empirically measured stage durations, with three contributions. First, the MITRE ATT&CK taxonomy is given a temporal layer, in which each stage carries a duration distribution estimated empirically from the observed episodes of that stage. Second, a probability-weighted multi-path formulation combines these durations with stage-transition probabilities to estimate the time remaining before the objective. Third, the framework is validated on a real multi-stage campaign rather than on synthetic traffic, and three survival models are compared under a matched protocol as a benchmark of how learnable the durations are. Random Survival Forest, DeepSurv, and DeepHit are compared on DAPT 2020, a public advanced-persistent-threat dataset of 82,577 real network flows collected across five days. Random Survival Forest reaches a stable concordance index of 0.92, with a standard deviation of 0.006 across twenty repeated stratified splits on leakage-free features, and it retains a concordance of 0.79 when benign traffic is excluded entirely. When the three models are placed on a single concordance scale and trained on an identical subsample of 40,000 flows, DeepSurv reaches 0.955 and DeepHit 0.879, so the neural models are competitive at that scale. DeepSurv nevertheless fails to converge on the full flow set, returning no survival estimates in any of five seeds, whereas the forest fits successfully at every training size examined. A stage-transition graph recovered from the data, built from 25 observed transitions across ten multi-stage sessions, shows branching progression, and the residual time, reported at the entry to each stage, falls along the campaign, from about 139 min at reconnaissance to about 31 min at lateral movement, conditional on reaching the objective. All stage-level estimates rest on 74 episodes from a single campaign, of which 43 carry a positive duration, so cross-environment generalisation remains to be confirmed. The framework gives a security operations centre a data-driven estimate of the active attack effort that remains before compromise, supporting informed containment decisions. Full article

Multi-branch networks combine backbones whose inductive biases are complementary, but their fusion modules set the branch weights from feature statistics alone and are therefore blind to what a pixel represents. We supervise fusion in label space instead. Ground-truth labels are converted into per-pixel routing targets, which are imposed on the branch weights through a Kullback–Leibler divergence term. Each pixel is thus routed towards the branch suited to its category: a convolutional branch for fine-grained boundaries, a state-space branch for large homogeneous regions, and a windowed-attention branch for intermediate-scale context. The mechanism is instantiated in HRFNet, a three-branch encoder with branch-specific dilation rates. HRFNet attains 76.85% and 79.88% mean intersection over union (mIoU) on Cityscapes and CamVid, averaged over five seeds, and 58.18% and 46.74% on the 59-class PASCAL Context and the 150-class ADE20K benchmarks, exceeding every baseline retrained under an identical budget. Ablations locate the gain in the routing rather than in added capacity: routed fusion adds 2.29% mIoU over average fusion, of which 1.53% comes from the label-space target itself, for 3.8M parameters and 8.8% of the network’s operations. All nine ablation contrasts remain significant after Holm–Bonferroni correction, and the gain persists in every two-branch configuration. Full article

Improving emergency vehicle mobility in congested urban environments is a critical challenge for transportation systems. Although roadway capacity expansions, such as widening roads, are often deployed to reduce congestion, their impact on emergency response performance is not always guaranteed, especially when delays concentrate at critical intersections. This study investigates how roadway capacity expansion affects emergency vehicle performance by using a microscopic traffic simulation framework. The study was applied to a real urban corridor in Mohammedia, Morocco, to provide a solid base for simulations with real-world conditions. A SUMO model was calibrated to represent two roadway configurations: a baseline two-lane layout and a three-lane post-widening scenario. Traffic volumes from 1056 to 3520 vehicles per hour were simulated, and performance was assessed using three emergency-specific indicators: Emergency Response Time (ERT), Delay Ratio (DR), and Priority Mobility Index (PMI). An initial single-run comparison suggested a substantial ERT reduction under moderate demand (343.40 s to 270.90 s, 21.11%); however, a 30-seed replication with paired Wilcoxon signed-rank tests shows that this and nearly all other widening effects are not statistically distinguishable from stochastic simulation noise. Only one of 12 emergency vehicle comparisons (Priority Mobility Index at 18:00) reached significance, and it favored the baseline configuration; none of 12 general traffic comparisons improved significantly, and general traffic was significantly slower under the widened configuration at 22:00 (p < 0.01). A supplementary sensitivity analysis (±20% emergency vehicle demand share) further shows that Delay Ratio conclusions are considerably more sensitive to this assumption (up to 34% relative change) than ERT or PMI (under 8%). These findings indicate that, in this network, roadway capacity expansion alone does not deliver a statistically robust improvement in either emergency vehicle or general mobility, and that a persistent signalized-intersection bottleneck remains the dominant constraint irrespective of lane geometry. The study provides a replicable, statistically validated simulation framework for assessing roadway capacity expansion effectiveness and cautions against single-run comparisons, which can substantially overstate the causal effect of infrastructure interventions in microscopic traffic simulation studies. Full article

Short-term GPU utilization forecasts are useful for scheduling, resource allocation, and capacity planning, but production traces are rarely smooth. They contain spikes, regime changes, idle periods, and incomplete observations. We study this problem on the MIT Supercloud Dataset using a direct, horizon-specific forecasting setup. After resampling the telemetry to 1 min intervals, the neural models are trained on min–max-normalized data and evaluated on the original 0–100% utilization scale after inverse transformation. Persistence and rolling mean predictors are added as non-trainable baselines and are evaluated on the same eligible targets as the neural models. Five sequence models—1D-CNN, GRU, FC-LSTM, Liquid Time-Constant Network (LTC), and Transformer—are compared at 1 min, 10 min, and 1 h horizons. To avoid a gross capacity imbalance, the primary model widths are chosen in a comparable range of approximately 55,000 trainable parameters. This controls the trainable model size only; the architectures still differ in computation, memory access, and optimization behavior. Besides the overall error, the experiments examine high-load periods, abrupt changes, and a 20% random zero-masking condition. Among the five neural models, FC-LSTM gives the lowest MAE and RMSE at the 1 min horizon. The persistence baseline reaches an MAE of 3.92 at this horizon, compared with 3.55 for FC-LSTM, corresponding to a 9.4% lower MAE for FC-LSTM. At 1 h, among the neural models, LTC has the lowest mean RMSE, while FC-LSTM retains the lowest MAE and WAPE. Paired GPU device-level comparisons indicate that the larger improvements over simple baselines are more robust than the small numerical gaps among the strongest neural models. The zero-masking robustness protocol is expanded to five masking seeds and all five primary neural models. Taken together, the results show that the preferred model changes with both the forecast horizon and error criterion. Full article

Seed transfer guidelines are used in ecological restoration to reduce the maladaptation risk from non-local genotypes. Researchers often examine second-generation populations under uniform conditions to isolate adaptation from environmental responses. However, in outcrossing species, interpreting these traits may be challenging if uncontrolled gene flow and recombination among provenances influence offspring phenotypes. We compared seed germination and seedling traits of the wind-pollinated grass Festuca vaginata, a dominant species of open sand grasslands in Hungary, across seed transfer zones (STZs) and localities. We used a two-generation common garden experiment with uncontrolled gene flow. Our results reveal that under common garden cultivation, the transgenerational stability of population differentiation is traitspecific. While locality effects on germination disappeared in the second generation, phenotypic variation in biomass and leaf length persisted, suggesting that shared environments homogenize germination faster than vegetative traits. As wind-pollinated species often exhibit weak regional genetic structuring, STZs may not capture the primary axis of phenotypic variation in specific traits. While not invalidating the use of STZs, our findings suggest that seed sourcing strategies should, if feasible, consider locality-level variation when using wind-pollinated grasses for species reintroduction. Full article

Economically motivated adulteration of olive oil, coffee and fruit juice is a persistent food-fraud problem for which Fourier-transform infrared (FTIR) spectroscopy with chemometrics offers rapid screening. Linear partial least squares (PLS) is interpretable but cannot capture non-linear mixing; neural networks add flexibility at the cost of becoming black boxes. We evaluated a Kolmogorov–Arnold network (KAN), which places learnable univariate functions on its edges and is therefore intrinsically interpretable, against PLS, support-vector regression, random forests, a multilayer perceptron and a one-dimensional convolutional network on three attenuated total reflectance (ATR)–FTIR datasets (olive oil + sunflower oil, coffee + malt flour, orange juice + apple juice; approximately 350, 400 and 400 spectra). All models were compared under identical, leakage-free validation that splits spectra by physical sample. The compact KAN was consistently competitive (cross-validated coefficients of determination (R²) = 0.86, 0.93 and 0.69) and yielded closed-form equations whose variables map to recognised vibrational bands and whose importance ranking agrees with SHapley Additive exPlanations (SHAP; Spearman ρ = 0.86–0.90); symbolic conversion costs no accuracy. We also report the following limits: PLS was strongest where the chemistry was linear (coffee) and the multilayer perceptron was strongest on fruit juice, whose equation is the weakest (R² = 0.47–0.75 across seeds); a parameter-matched perceptron matched the KAN’s accuracy; and leave-one-brand-out validation degraded every model. The KAN is therefore a promising, compact and genuinely transparent alternative under controlled multi-matrix conditions, not a deployment-ready method. Full article

Delayed post-sowing watering may leave treated wheat (Triticum aestivum) seeds in dry soil for several weeks before germination begins, raising the question of how long the effects of seed-applied biostimulants remain biologically active under these conditions. This study evaluated the persistence of the biological effects of a seed-applied biostimulant, applied alone (BS) or combined with a triticonazole-based fungicide (TBS), after delayed watering of wheat seeds. Seeds were watered immediately after sowing or after 2, 3, 4, or 5 weeks and evaluated in Petri dishes and growth pots. Compared with the untreated control, BS and TBS reduced mean germination time by up to 9.4 h and 8.7 h, respectively (p < 0.05), while also promoting greater seminal root development after prolonged exposure to dry conditions. In the pot experiment, treatments including the biostimulant maintained greater root dry weight at both BBCH 11 and BBCH 12, indicating that a measurable stimulatory response was still observed beyond germination and early seedling establishment. The strongest responses were observed when watering was delayed by 2–3 weeks, remained evident after 4 weeks, and declined after 5 weeks of dry sowing conditions. Under the conditions of this experiment, the results demonstrate that, although the biostimulant supports early root system development after moisture restoration, the magnitude of the treatment response is conditioned by the duration of the pre-germination waiting period. Full article

A fixed-odds contract on short-horizon price direction has a positive expected value only when its win probability exceeds the break-even rate implied by the payout ratio. A deployable predictor must also produce signals at a sufficiently stable rate. We formulate this setting as selective prediction with a coverage constraint and combine a five-seed gradient-boosting ensemble over a 90-dimensional causal feature panel with daily adaptive quantile thresholds, each estimated from the preceding 14 to 28 days of model scores, with parameters selected on training data alone. Configurations are frozen after three chronological pseudo-out-of-sample folds and evaluated on a held-out period from 1 January to 10 June 2026, and the whole procedure is then repeated on a quarterly re-freezing cadence over seven successive windows. Across BTC and ETH at 5- and 10-min horizons, with a payout of 0.8 and a 55.56% break-even rate, the models execute 10.4 to 11.0 trades per day, and all four selective win rates exceed break-even. Under a dependence-aware block bootstrap, three of four remain significant, and within a 32-test confirmatory family, two survive Holm–Bonferroni correction. Coverage stays inside the operational band in 26 of 28 re-frozen windows. Compared under one execution protocol, a fixed calibration slice drifts out of band while a trailing window does not, and adaptive conformal inference (ACI) matches the proposed rule on coverage when its step size is tuned but not otherwise, whereas an outcome-driven conformal controller reduces coverage by more than an order of magnitude. The expected value is insensitive to exchange fees, which consume under 5% of the measured edge, and sensitive to the payout term. Under matched feature sets, training pools, and coverage, most of the apparent cross-asset difference does not persist. This paper presents a proof of concept for the framework rather than making any claim about cryptocurrency predictability. Full article

Global efforts to end hunger are about more than producing enough food; they are also about producing food that is nutritious enough. Micronutrient deficiencies—the “hidden hunger” affecting billions—persist in part because the domestication bottleneck quietly eroded mineral and lipid diversity from the very crops the world relies on most. Wild relatives of domesticated legumes still carry much of that diversity, and genetic biofortification offers a sustainable route to put it back to work. Wild Vigna germplasm remains poorly characterized for traits that could support nutritional biofortification and neo-domestication. With that in mind, we characterized the seed mineral and fatty acid composition of 86 accessions from four wild Vigna species (V. vexillata, V. ambacensis, V. reticulata and V. racemosa), benchmarked against three domesticated (cultivars) and semi-domesticated checks (V. unguiculata, V. umbellata and V. vexillata landrace). Copper, manganese, zinc and iron were quantified by flame atomic absorption spectrophotometry after dry-ash digestion, and fatty acids were profiled as methyl esters by GC-MS. The species differed systematically in their mineral profiles. V. reticulata stood out as the most promising donor for copper-focused breeding, V. vexillata carried the highest median Zn, Mn and Fe values and is attractive for multi-micronutrient improvement, V. ambacensis showed a more stable but less extreme profile, and V. racemosa combined a relatively high Fe concentration with the most nutritionally favorable lipid profile of all—dominated by the essential polyunsaturated linoleic (C18:2n 6) and α-linolenic (C18:3n 3) acids. The other three wild species, by contrast, were dominated by saturated palmitic (C16:0) and stearic (C18:0) acids, which gives their oils greater oxidative stability and different food-industry applications. Principal component analysis supported these patterns for both datasets—minerals (PC1 = 60.39%, PC2 = 20.21%; cumulative 80.61%) and fatty acids (PC1 = 81.6%, PC2 = 14.0%; cumulative 95.6%)—and cleanly separated V. racemosa and the checks from the remaining wild species on lipid composition. Together, these results identify concrete targets for marker-assisted biofortification and de novo domestication of four African Vigna taxa. Full article

Heavy metal deposition along high-traffic roadsides poses a persistent threat to agricultural safety, yet the partition barrier efficiency across rhizosphere–root–shoot interfaces in perennial nut crops remains poorly understood, representing a significant research gap. This study determined the concentrations of potentially toxic elements in the rhizosphere soils and distinct organs (leaves, pericarp, and edible seeds) of Siirt pistachio trees along a distance gradient (0, 50, and 100 m, plus a control site) in the Siirt and Tillo districts. To filter analytical baseline noise, all raw datasets were subjected to strict solid-matrix limit of detection (LOD) screening using a standardized dilution factor of 30 mL/g (DF = 15 mL final volume/0.5 g sample mass). Soil analysis revealed that the alkaline pH (6.90–7.27) and highly calcareous nature (21.97–65.75%) of the rhizosphere acted as a powerful edaphic barrier, immobilizing metals in the soil and limiting their translocation to aboveground tissues. Plant accumulation followed a leaf > pericarp > seed hierarchy, proving the canopy’s role as an effective vegetative filter. Crucially for food safety, highly toxic Cd (<1.74 µg/kg) and Bi remained entirely below detection limits in edible seeds. Cr peaked in leaves (730.42–795.00 µg/kg) but was highly restricted in seeds. Detected kernel concentrations of As, Co, Ni, Pb, and Sb were strictly below international toxic thresholds, while essential Cu physiologically concentrated in seeds and leaves. Consequently, the cumulative Hazard Index (HI) remained exceptionally below the 1.0 critical safety limit for both adults (<0.18) and children (<0.32). This confirms that roadside pistachios pose zero non-carcinogenic health hazards and are completely safe for human consumption. Full article

This study proposes MSATE-Net for next-day stock index forecasting. The model combines parallel one-dimensional convolutions with receptive fields of 3, 7, and 15 trading days, a bidirectional LSTM operating entirely inside a historical lookback window, sample-dependent temporal attention, and a residual regularized prediction head. Here, “bidirectional” denotes paired processing of the same observed window; it does not assert time-reversal invariance of financial prices or access to observations after the forecast origin. The globally learned attention temperature controls overall selectivity and is not described as a regime-specific adaptive parameter. Experiments use S&P 500, CSI 300, and Nikkei 225 data; persistence and drift benchmarks; recent forecasting architectures; five-seed uncertainty estimates; expanding-window tests; return and directional metrics; and Diebold–Mariano comparisons. The revised evidence supports lower price-level errors, while directional and significance results are mixed across markets. Because a separate model is fitted in each market, the findings establish cross-market consistency rather than transfer learning. Full article