Search Results (1,199)

Automatic modulation classification (AMC) is a critical enabler of cognitive radio (CR) systems, allowing secondary users to identify primary user modulation schemes and adapt transmission parameters in real time. Traditional AMC approaches, based on likelihood functions or hand-crafted features, suffer from degraded performance under low signal-to-noise ratio (SNR) conditions and realistic channel impairments. In this paper, we propose MS-SENet (Multi-Scale Squeeze–Excitation Network), a novel deep-learning architecture that integrates multi-scale convolutional feature extraction, squeeze-and-excitation channel attention, residual learning, bidirectional long short-term memory (BiLSTM) temporal modelling, and global attention pooling into a unified framework for robust AMC. The multi-scale convolution module employs parallel branches with kernel sizes of 3, 5, and 7 to capture both fine-grained phase transitions and coarse envelope patterns from raw in-phase/quadrature (I/Q) signal samples. Squeeze–excitation residual blocks perform channel-wise feature recalibration, enabling the network to emphasize informative feature maps while suppressing less relevant ones. A bidirectional LSTM layer models temporal dependencies across the signal sequence, and a global attention pooling mechanism performs weighted temporal aggregation prior to classification. We present a comprehensive taxonomy of deep-learning architectures for AMC organised along five axes—input representation, feature extraction, temporal modelling, regularization strategy, and architectural complexity—and conduct a rigorous comparative evaluation against ten baseline architectures on a RadioML-style synthetic dataset (110,000 samples, 11 modulation classes, and 20 SNR levels from −20 to +18 dB). The experimental results demonstrate that MS-SENet achieves a mean classification accuracy of 87.9% at SNR ≥ 0 dB (the average of the medium and high SNR regime averages: 86.06% for 0 ≤ SNR < 10 dB and 89.68% for SNR ≥ 10 dB) while maintaining a compact footprint of approximately 406 K parameters, making it suitable for deployment on resource-constrained edge devices. We further analyze the robustness of the proposed architecture to multipath fading, carrier frequency offset, and sample rate offset, confirming its resilience under practical operating conditions. MS-SENet is an architecture designed for automatic modulation classification of I/Q signals and is not related to the homonymous architecture for speech emotion recognition. Full article

Soybean seeds have long been regarded as “storehouses of high-quality proteins”. The breakdown of dietary proteins by digestive proteases is essential for achieving adequate protein digestibility in animals and humans. However, plants have evolved a diverse array of protease inhibitors that regulate or restrict protease activity. In soybean, these inhibitors are concentrated primarily within the 2S protein fraction. Trypsin inhibitors (TIs) of Kunitz trypsin inhibitor (KTI) and Bowman–Birk inhibitor (BBI) are the most impactful due to their strong anti-tryptic activity, which interferes with digestive proteases in humans and animals. Elevated TI levels render raw soybeans unsuitable for direct food or feed use unless thermal or processing inactivation treatments are applied. Elimination or reduction in KTI and BBI using classical and biotechnology-based breeding efforts is a promising strategy. Soybean germplasm harboring BBI null alleles has not been reported. Breeding only for low or null KTI content in soybean would not be sufficient for practical applications. Hybridizing IT105782 × PI 547656 and using the reported Kompetitive Allele-Specific PCR (KASP) markers represents an effective classical breeding strategy. Simultaneous CRISPR/Cas9-mediated knockout of key KTI and BBI genes is expected to enable the development of soybean lines with substantially reduced TI levels, an outcome that cannot be readily achieved through classical introgression of null alleles, as naturally occurring null BBI alleles have not yet been identified. Moreover, this approach avoids the linkage drag associated with donor-derived null KTI alleles. However, this approach remains challenging due to functional redundancy and compensatory effects among KTI and BBI family members, extensive sequence homology among KTI and BBI genes that complicates the minimization of off-target effects, and the genotype dependency of Agrobacterium-mediated soybean transformation. Microtiter plate AACCI/AOCS could be one practical option for measuring TIA in breeding programs in terms of precision. Potential trade-offs associated with reduced trypsin inhibitor levels, including possible effects on plant defense and stress resistance, should be investigated in future studies, as these aspects have received little attention in previous research. Full article

HEV genotype 3 is classified into several subtypes. In Italy, 3f, 3c, 3e and, rarely, 3a subtypes are usually detected in human hepatitis E cases: the present study documents, for the first time, the detection of the 3i subtype, previously described in Italy exclusively from wild boars. Routine surveillance by short Sanger sequences highlighted five hepatitis E cases with unusual HEV subtype between 2019 and 2023 in a small geographical area in Umbria, Central Italy. Further characterization of the whole HEV genome by NGS was successful in three of them. Through phylogenetic and p-distance analysis, all the sequences could be classified as subtype 3i, and three of them proved to be highly related to some strains observed in wild boars sampled in the same geographical area. The 3i subtype had never been detected previously in humans in Italy; it is likely that increased local circulation of 3i-like strains in wild boars may have increased the chance for their transmission to humans. Investigation of the possible risk factors for HEV infection highlighted consumption of raw/undercooked products from wild boars/pigs in three cases. However, in the remaining two cases, this source seems unlikely, and transmission might have occurred indirectly from contaminated sources, such as wastewater and home-grown vegetables. Further studies are needed to investigate if these latter sources may be more common than previously thought or may play a role in relation to specific HEV subtypes and/or only in rare and exceptional circumstances. Full article

The COVID-19 pandemic highlighted the urgent need to develop effective and broad-spectrum antiviral therapies against coronaviruses. One strategy to address this concern is a combination therapy using repurposed drugs against zoonotic viruses with pandemic potential. We previously demonstrated that the combination of Remdesivir and Ivermectin is highly potent and synergistic in inhibiting the replication of murine hepatitis virus (MHV) in RAW264.7 macrophages. This study investigated the interactions between the drug combination, coronavirus and host by proteomics and RNA sequencing of MHV-infected H2.35 murine liver epithelial cells. Time-of-addition and time-of-removal assays suggested that the drug combination likely affected the synthesis of viral RNA and viral protein. This combination drastically diminished the live virus titer greater than the respective monotherapies in MHV-infected H2.35 cells (by ~4 log₁₀), as well as in SARS-CoV-2-infected VeroE6 cells and human nasal epithelial cells. Proteomic and transcriptomic analyses revealed that viral protein and RNA levels were significantly depressed upon combination treatment. The drug combination exhibited considerable negative effects upon host RNA processes and resulted in the upregulation of host protein processes (e.g., response to unfolded protein; protein insertion into ER membrane). Molecular pathways affected by the combination treatment were markedly distinct from the monotherapies and indicated that Ivermectin enhances Remdesivir by modulating critical host processes to synergistically exert its inhibitory effect on the coronavirus replication cycle. Full article

Background/Objective: Bisphenol AF (BPAF), a fluorinated analogue of bisphenol A, is an environmental contaminant associated with hepatotoxicity and metabolic disruption. However, the systematic molecular mechanisms linking early transcriptional events to metabolic dysfunction in the liver remain poorly defined. The aim of this study is to elucidate the association between BPAF exposure and hepatic lipid accumulation by integrating transcriptomics, cellular metabolomics, and targeted phenotypic assays. Methods: We performed RNA-sequencing on livers from mice exposed to BPAF (0.1–10 mg/kg/day, 28 days), and performed non-targeted metabolomics on AML12 murine hepatocytes co-cultured with RAW264.7 macrophages in a Transwell system (0–2500 nM BPAF, 48 h). Key metabolic pathways were identified through integrated bioinformatics and validated using enzymatic assays, qRT-PCR, Western blotting, and phenotypic staining (lipid droplets, ROS). Results: Multi-omics integration revealed significant disruption of PPAR signaling and the tricarboxylic acid (TCA) cycle. A striking dose-dependent accumulation of succinate was observed in exposed cells, concomitant with a significant inhibition of succinate dehydrogenase (SDH) activity (52% reduction at 2500 nM, p < 0.001). Transcriptomic data confirmed the downregulation of mitochondrial fatty acid β-oxidation genes. Phenotypic validation indicated that BPAF exposure is associated with oxidative stress, pro-inflammatory cytokine release (TNF-α, IL-6), and pronounced intracellular lipid droplet accumulation in hepatocytes. Conclusions: This study suggests that BPAF exposure is associated with SDH dysfunction, TCA cycle arrest, and lipid dysregulation. Whether BPAF directly inhibits SDH or acts through upstream mitochondrial targets warrants further structural and kinetic investigation. Full article

Traditional cheese products harbor complex microbial communities that influence their quality and safety. However, the effects of processing scale and manufacturing stage on the microbial profile of hurood, a traditional Mongolian cheese, remain poorly understood. This study examined microbial indicators, community composition, and succession dynamics across four manufacturing stages (raw milk, yogurt, whey, and hurood) and three processing scales (pastoral household, workshop, and factory) using plate counting and 16S rRNA gene amplicon sequencing. Twenty-four samples were collected from Xilin Gol, Inner Mongolia. Total aerobic plate counts and coliform counts decreased significantly from raw milk (7.30 and 4.49 log CFU/g, respectively) to hurood (2.02 and 0.34 log CFU/g, respectively; p < 0.05), reflecting progressive microbial reduction through acidification and thermal treatment, whereas yeast counts remained stable across stages. Firmicutes dominated the fermented stages, with Lactococcus and Lactobacillus as the predominant genera. Whey harbored an exceptionally high abundance of Acetobacter (21.6%), highlighting its valorization potential. Factory-scale production yielded the lowest mold and coliform counts in finished products despite higher initial coliform levels in industrial raw milk, reflecting the effectiveness of standardized hygiene management. In contrast, workshop-scale samples exhibited a higher relative abundance of environmental indicator bacteria, suggesting a comparatively elevated contamination risk this intermediate production scale. PICRUSt2-based functional predictions indicated stage-specific metabolic potential, including predicted enrichment of pyruvate and fatty acid metabolism in yogurt, amino acid metabolism in whey, and vitamin B6 metabolism in hurood. These findings provide a systematic microbial baseline for hurood, identify scale-specific microbiological risk profiles, and offer a foundation for targeted hygiene control and standardized production strategies. Full article

DoS and DDoS attacks remain a major threat to service availability in modern IP and IoT networks, yet many learning-based detectors depend on dataset-specific flow exports, feature tables, or preprocessing conventions. This article presents a unified sequence-level detection pipeline designed to process heterogeneous public datasets through the same representation. Raw PCAP/PCAPNG traces from CIC-IDS-2017, CIC-DDoS-2019, and CICIoT2023 are converted into one-second aggregates per destination host using header-only features derived from IP, TCP, UDP, and ICMP metadata, source diversity, and packet timing. Dataset-specific annotations are used only to assign binary DoS/DDoS labels to this common representation. The resulting time-ordered aggregates are grouped into fixed-length temporal windows and classified by a compact transformer encoder, TemporalDosTransformer, which produces a window-level attack probability. The study focuses on whether a clean PCAP-based aggregation and labelling flow can support consistent DoS/DDoS detection across multiple datasets without payload inspection, flow-exporter dependence, or dataset-specific feature engineering. Full article

The increasing consumption of fresh organic produce has given rise to concerns regarding the microbiological safety of minimally processed foods. Organic cultivation may be associated with increased exposure to environmental microorganisms due to soil-based inputs and reduced chemical interventions, including both beneficial taxa and potential foodborne pathogens. Fresh produce is known to harbour complex microbial ecosystems, which are shaped by farming practices, plant physiology, handling, packaging and storage, particularly in raw-consumed products such as leafy greens and strawberries. In this study, bacterial (16S rRNA) and eukaryotic (18S rRNA) communities were characterized by amplicon sequencing. In parallel, an amoeba-associated bacterial microbiome was analyzed and DVC-FISH was used to assess the viability and metabolic activity of pathogenic bacteria internalized within free-living amoebae (FLA). No significant differences in alpha or beta diversity were observed between organic and conventional products, suggesting microbiome convergence at the retail stage driven by post-harvest handling and processing. Potentially pathogenic genera, including Pseudomonas, Stenotrophomonas, and Acinetobacter (bacterial), as well as Tilletiopsis, Candida, and Naegleria (eukaryotic), were identified in both organic and non-organic microbiomes. The viability of FLA-internalized Pseudomonas spp. was confirmed by DVC-FISH, demonstrating that FLA act as reservoirs, enhancing pathogen persistence in fresh produce. This integrated assessment of organic and conventional fruits and vegetables at the retail stage highlights the importance of post-harvest handling and retail conditions in shaping microbiological safety. The integration of microbiome profiling with targeted viability analyses demonstrates that downstream stages are critical control points for food safety and consumer exposure, beyond the influence of the production system alone. Full article

Jute (Corchorus spp.) is the most important bast fiber crop, providing raw materials for textiles, bio-composites, and papermaking. This study analyzed the chloroplast genomes of two wild long-fruited jute species: Qiaojianyehuangma (QJYHM) and Maliyehuangma (MLYHM). The chloroplast genomes exhibited typical circular quadripartite structures (LSC, SSC, IRa/IRb), containing 129 genes (37 tRNA, 8 rRNA, 84 mRNA). Overall GC content was 36.76%, indicating high genetic conservation. Compared with cultivated varieties, wild varieties exhibit differences in LSC region length, IR boundary positions, and repetitive sequences, reflecting minor sequence variations in the chloroplast genome that occurred during domestication. Codon preference analysis showed both wild species favor A/U-ending synonymous codons, with a strong preference for methionine’s AUG codon. Repetitive sequence analysis revealed 280 and 252 dispersed repeats in Qiaojianyehuangma and Maliyehuangma, respectively, primarily mononucleotide SSRs. Based on Ka/Ks analysis, it was discovered that most chloroplast genes were under purifying selection. In contrast, positive selection signals were detected in rpl23, ycf1, and ycf2, implying their involvement in adaptive evolution. We identified 161 polymorphic sites (97 SNPs, 64 InDels), with ycf1 as a mutation hotspot. Phylogenetic analysis clustered both wild species with Corchorus capsularis with a 100% bootstrap value, forming a well-supported sister group. This study provides basic chloroplast genome data for two wild Corchorus olitorius accessions, revealing their conserved genomic features and minor sequence variations. Full article

Soybean is an important food and oil crop, and rapid nondestructive identification of seed cultivars is crucial for seed purity inspection, varietal certification, breeding management and food-quality control. However, the global spectral profiles of individual soybean seeds from different cultivars are often highly similar, making it difficult for single-representation models to simultaneously capture spectral sequential dependency and inter-band relational structure. To address this issue, this study proposes a GADF–Mamba Fusion Network (GMFNet) for soybean seed hyperspectral classification. Hyperspectral images of 24,800 seeds from eight cultivars were acquired, and reflectance spectra in the range of 900–1700 nm were collected. After preprocessing, 200 effective bands were retained. The preprocessed one-dimensional spectral sequence was fed into a Mamba-based branch to model continuous wavelength dependency and global spectral evolution, while the same sequence was transformed into a GADF image, resized to 208 × 208, and input into a ResNet18-based structural branch to extract inter-band relational features. The two heterogeneous representations were then integrated through a weighted feature fusion module for final classification. Experimental results showed that Mamba achieved the best test accuracy (0.8721) among the raw spectral models, whereas ResNet18 achieved the best test accuracy (0.8737) among the GADF-based structural models. More importantly, the proposed weighted fusion strategy achieved the best overall performance, reaching validation and test accuracies of 0.9039 and 0.9011, respectively. These results demonstrate that spectral sequential information and GADF-based structural semantics are highly complementary. Overall, the proposed framework provides an effective hyperspectral solution for single-seed soybean cultivar identification and shows potential for non-destructive automated quality control in food-industry applications. Full article

Accelerometer sensors and artificial intelligence (AI) are reshaping automated behavior monitoring in precision livestock management, yet their joint deployment on extensive rangelands is constrained by energy and bandwidth budgets. Low-Power Long-Range Wide-Area Network (LoRaWAN) collars address these constraints by compressing the raw tri-axial signal on the device into a single scalar per reporting interval, the Motion Index (MI). This onboard compression preserves enough signal to separate active behaviors but discards the per-axis and frequency content that fine-grained classification typically relies on. On a dataset of 9222 labeled observations from 24 cows across four breeds, MI distinguishes walking from grazing reliably but fails to separate ruminating from resting; both correspond to a stationary animal and yield near-zero, statistically indistinguishable distributions. Earlier MI-only models reached only about 65% four-class accuracy, and ruminating was commonly merged into resting. We show that much of this loss can be recovered by treating the MI stream as a time series. Session-aware lag features, rolling statistics, and an autoregressive previous-behavior feature lift four-class macro-F1 from 0.647 to 0.94, with per-class F1 of 0.95 for ruminating and 0.92 for resting (and at least 0.92 for every behavior). In autonomous deployment the previous behavior must be predicted rather than observed; for this setting we add a Viterbi sequence-decoding step that combines the classifier’s per-step outputs with a learned behavior-transition model, recovering a substantial part of the ruminating signal from the activity stream alone while keeping walking and grazing reliable. The gain is consistent across seven classifiers and four genetically distinct breeds, indicating that it is driven by the features rather than by a specific model. Full article

High-entropy MAX phases (TiVNbTa)₂AlC have attracted increasing attention due to their potential advantages in structural stability, damage tolerance, and mechanical reliability under complex service environments. This work studied the crystal and electrical structures with the elastic properties, the synthesis reactions and further wear resistance of HE-MAX (TiVNbTa)₂AlC theoretically and experimentally. The charge transfer between both M-C atoms and M-Al atoms turned more intense, which correspondingly strengthened the M-C and M-Al bonds, respectively. Because of the dope on M-sites, (TiVNbTa)₂AlC exhibited larger fracture toughness K_IC and a lower brittle index M, which suggested lower brittleness, better crack extension resistance, and higher damage tolerance than Ti₂AlC. In this work, high-entropy (TiVNbTa)₂AlC MAX phase ceramics were successfully synthesized by a powder metallurgy route combined with pressureless sintering and spark plasma sintering (SPS). The effects of raw material composition and sintering temperature on phase evolution, microstructure formation, mechanical properties, and tribological behavior were systematically investigated. The results show that a highly pure (TiVNbTa)₂AlC phase with a phase fraction of 96.8% could be obtained at a molar ratio of M:Al:C = 2:1.2:0.8 and a sintering temperature of 1550 °C. Phase evolution analysis indicated that the reaction process followed the sequence of intermetallic compound (IMC) formation → carbide formation → MAX phase formation. Severe lattice distortion induced by the multi-principal-element solid solution significantly enhanced the hardness of the material, which was markedly higher than that of conventional ternary MAX phases. Owing to its higher hardness and more homogeneous solid-solution structure, HE-MAX (TiVNbTa)₂AlC could inhibit the formation of surface microcracks and reduce the driving force for crack propagation to some extent. Therefore, the lower wear rate not only reflected improved tribological performance but also demonstrated the beneficial role of high-entropy design in enhancing resistance to surface damage. Full article

Autonomous navigation in unstructured off-road environments presents fundamental challenges due to terrain heterogeneity, the absence of structured road markings, and the necessity for real-time traversability reasoning from raw sensory observations. We present TerrainFormer, a hierarchical framework that integrates a world model for terrain dynamics prediction with a temporal decision transformer for action selection. Our methodology employs a two-phase training paradigm: (1) self-supervised world model pretraining on LiDAR point clouds to learn terrain representations encompassing traversability, elevation, and semantic segmentation; (2) behavioral cloning of the decision transformer conditioned on frozen world model features with temporally derived goal directions. The world model processes raw 3D LiDAR point clouds through a PointPillars encoder for real-time bird’s-eye-view (BEV) projection, followed by a Vision Transformer backbone that produces latent terrain representations. A principal contribution is our cross-dataset generalization paradigm: the world model is trained on separate datasets while the decision transformer is trained on separate sequences, ensuring zero data overlap between training phases. We introduce automatic goal direction computation from vehicle pose trajectories, enabling the model to learn directionally conditioned navigation policies. To address the class imbalance inherent in off-road driving data, we employ focal loss with inverse-frequency class weighting and action-chunk supervision. Experimental evaluation on the RELLIS-3D dataset achieves 87.31% test accuracy with 0.7948 macro F1 across all 12 action classes. The world model’s predicted future frames produce only a 0.79% accuracy drop versus ground-truth observations, with 98.82% action agreement, demonstrating effective cross-dataset generalization for real-time off-road navigation. Full article

Background/Objectives: Traditional protein contact analysis often fails to distinguish between local, sequence-driven motifs and global, tertiary scaffolding, which ensures structural determinism. While deep-learning models do not fully elucidate the ‘why’, they do reveal the underlying directional rules of the stability landscape. In this study, we analyzed 475 non-redundant Protein Data Bank (PDB) structures categorized into SCOP classes (all-α, all-β, α/β, α+β) of the hydrolase superfamily. Methods: To isolate the structural anchors of the global fold, we applied a sequence separation filter of ∣i − j∣ ≥ 6 and a precise spatial cutoff of 3–5 Å between Cα-only to construct asymmetric 20 × 20 frequency matrices, both raw and normalized, then present the former using a violin diagram. We developed a Pearson Correlation (PC) framework to analyze these matrices, providing high correlation when considered as vectors and giving the directionality (N-to-C vs. C-to-N) in protein folding when considered as matrices. Results: Our results reveal a hierarchical organization of tertiary determinism. Initial visualization of Residue–Residue Contact Frequency Matrices (RRCFMs), Z-score normalization (NRRCFM), and violin plots reveal the Universal Structural Grammar (USG) of interaction. Furthermore, a near-perfect PC (r = 0.99) as determined via inter-class Z-score correlation and inter-class PC demonstrates shared statistical interaction laws. In addition, PC Stage 1 (intra-class) analysis identified high symmetry, with around 80% of contacts exhibiting a very strong to strong positive correlations, while PC Stage 2 (inter-class) analysis demonstrated that around 50% of contacts exhibited very strong to strong positive correlations. Finally, we identified universal druggable pockets for drug discovery. Conclusions: This powerful mathematical framework provides a robust analytical tool for structure-based drug design. Full article

Drone-to-drone (D2D) detection is a critical yet underexplored task in low-altitude intelligent perception, where UAV targets are often small, weakly textured, motion-affected, and disturbed by complex backgrounds and environmental changes. Existing cross-domain detection datasets mainly focus on ground objects or single-factor shifts, making them insufficient for evaluating D2D detection under coupled real-world variations. To address this gap, we present TriCross-D2D, an RGB air-to-air UAV detection dataset and benchmark with three explicit domain shifts: scene, viewpoint, and weather. Built from real flight videos and controlled synthetic fog, TriCross-D2D contains 13 RGB video sequences, 23,403 raw frames, 7045 benchmark images, and 9771 annotated UAV instances. It provides a fixed split of 4045 Source_train images, 2000 Target_train images, and 1000 Target_val images, supporting both unsupervised domain adaptation (UDA) and semi-supervised domain adaptation (SSDA). The dataset is dominated by small objects, with extremely tiny, tiny, and small targets accounting for 73.8% of all instances. Benchmark results show that existing cross-domain detectors still perform limitedly on TriCross-D2D, especially under stricter localization and recall metrics. Single-factor analysis further reveals that the coupled scene–viewpoint–weather protocol is more challenging than isolated shifts, with viewpoint variation producing a particularly strong domain gap. As an exploratory enhanced baseline, SCOPE-DA-RTDETR improves DA-RTDETR from 28.63/13.12/22.39 to 29.94/13.71/23.40 in ${\mathrm{AP}}_{50}/{\mathrm{AP}}_{50–95}/\mathrm{AR}$, showing consistent but modest gains. These findings demonstrate that TriCross-D2D provides a challenging and discriminative benchmark for cross-domain D2D small-object detection. Full article