Search Results (228)

Hollow-core anti-resonant fibers (HC-ARFs) have emerged as a promising platform for next-generation optical systems, offering attractive advantages in low-latency, low-nonlinearity, and high-power handling. However, the development of high-performance functional components, such as polarization beam splitters (PBSs), within this platform faces a significant challenge: the simultaneous achievement of ultra-broad bandwidth, compact device length, high polarization selectivity, and strict single-mode operation remains elusive. To address this challenge, we propose and numerically investigate a novel dual hollow-core anti-resonant fiber (DHC-ARF) based on a hybrid nodal–nodeless architecture. The design integrates three functional units: (1) an asymmetric nested semi-elliptical tube pair that defines the dual cores and serves as the primary wavelength-insensitive coupling channel; (2) nodeless nested circular tubes positioned peripherally to effectively suppress higher-order mode propagation while maintaining low fundamental mode loss; and (3) a selective localized thick-wall region that introduces a polarization-dependent perturbation to the x-polarized supermodes, whose observed behavior is physically consistent with a phase-mismatch effect associated with anti-crossing-like modal interaction near the target wavelength. Through synergistic optimization of these elements, we numerically demonstrate a combination of performance metrics. At the central wavelength of 1.55 µm, the coupling length for the y-polarization (Lcy) is reduced to 6.35 cm, while the coupling length ratio (CLR = $L_{\mathrm{cx}}$/$L_{cy}$) equals 2.001, indicating effective polarization selectivity. Consequently, a device length of 12.7 cm is numerically demonstrated, which is comparable to or shorter than existing ultra-broadband DHC-ARF PBS designs. The proposed PBS is numerically shown to exhibit an ultra-broad bandwidth of 460 nm (spanning 1320 to 1780 nm) with a polarization extinction ratio better than 20 dB, peaking at 53 dB. Furthermore, HOMER (λ) remains above 100 throughout the operating band and exceeds 200 over most of the band, indicating robust single-mode operation. This work not only presents a PBS design with competitive overall performance but also provides a versatile structural paradigm for developing functional components in hollow-core fiber-based integrated optical systems for high-speed communications and precision sensing. It should be noted that this work is based on numerical simulations, and experimental fabrication and validation will be pursued in future work. Full article

Cross-reactions between peroxy radicals (RO₂) derived from different volatile organic compound (VOC) precursors have been proposed as an important pathway during atmospheric oxidation. However, direct molecular evidence has been limited. In this study, α-pinene and fully deuterated toluene (d8-toluene) were oxidized separately and as a mixture in a potential aerosol mass (PAM) flow reactor, and the resulting secondary organic aerosol (SOA) was characterized by a high-resolution mass spectrometer (ESI FT-ICR-MS). A constrained chemical mass balance (CMB) model attributed 82.9% of the mixed-SOA signal to single-precursor sources (66.5% α-pinene, 16.4% d8-toluene), leaving a 17.1% signal-based residual fraction unexplained by linear mixing. Among 2450 residual molecular formulas exclusive to the mixed-SOA, 1858 were identified as cross-reaction candidates, with carbon, oxygen, and double bond equivalents (DBE) distributions consistent with RO₂-RO₂ cross-reactions between toluene- and α-pinene-derived fragments. We also identified representative monomer-dimer pairs, where one monomer corresponded to a known α-pinene oxidation product, while the other matched a primary oxidation product of d8-toluene oxidation based on the Master Chemical Mechanism (MCM), providing strong molecular-level evidence for RO₂-RO₂ cross-reactions. Our findings demonstrate that the mixed VOCs generate unique SOA products that extend beyond simple additive chemistry, with implications for SOA yield parameterizations and chemical transport models. Full article

Structured radiology reporting can improve clinical decision support by standardizing clinical findings into hierarchical formats. However, thousands of questions in structured report templates about clinical findings are prohibitively time-consuming, which can limit clinical adoption. Furthermore, early medical VQA datasets primarily focused on free-text and independent question–answer pairs while a recent dataset, Rad-ReStruct, introduced a hierarchical VQA, but the accompanying model still relies heavily on flattened embedding representations and single-path text–image fusion mechanisms that inadequately handle complex hierarchical dependencies in responses. In this paper, we propose DPA-HiVQA (Dual-Path Cross-Attention for Hierarchical VQA), addressing these limitations through two key contributions: (1) multi-scale image embedding representing global semantic embeddings with patch-level spatial features from domain-specific BioViL encoder; (2) dual-path cross-attention mechanism enabling simultaneous holistic semantic understanding and fine-grained spatial reasoning. Evaluated on the Rad-ReStruct benchmark, the model substantially outperforms the established benchmark baseline with an overall F1-score and Level 3 F1-score improvement by 21.2% and 31.9%, respectively. The proposed model demonstrates that dual-path cross-attention architectures can effectively connect holistic semantic understanding and fine-grained spatial detail, paving the way for practical AI-assisted structured reporting systems that reduce radiologist burden while maintaining diagnostic accuracy. Full article

Multi-link terrestrial backhauls create a shared transport environment in which heterogeneous multipath protocols compete for the same forwarding resources while reacting to congestion with different control logics. In this paper, we investigate this problem in a 5G Integrated Access and Backhaul (IAB) scenario where an IAB node aggregates traffic from multiple User Equipments (UEs) and forwards it toward the core network over two terrestrial backhaul paths. We focus on the coexistence of Multipath TCP (MPTCP) and Multipath QUIC (MPQUIC), evaluating how cross-protocol Congestion Control (CC) pairings affect performance. Specifically, all feasible BBR, CUBIC, and Reno cross-pairings are assessed under symmetric and asymmetric dual-backhaul conditions, considering Enhanced Mobile Broadband (eMBB) and dense low-rate traffic regimes representative of mMTC-like operation. The analysis considers throughput, Jain’s fairness index, jitter , and packet loss to identify the trade-offs of each CC pairing. Results show that CC selection is a first-order design factor in MPTCP/MPQUIC coexistence over shared backhauls. No single pairing is uniformly optimal across all metrics: some configurations provide more balanced throughput sharing, others improve fairness, while the most favorable solutions for jitter do not necessarily maximize transport efficiency. These findings identify CC pairing as a tuning dimension for multi-link backhaul systems based on heterogeneous multipath transports. Full article

In the context of political and financial market turmoil, effectively forecasting financial market trends is crucial for investment decisions. Large language models (LLMs) have been applied in extant research to predict market trends, analyze investor sentiments and interpret financial news, all aiming to help investment decision making. However, LLMs face limitations due to training data heterogeneity, restricting multidimensional perspectives and hindering comparative analysis for optimization. This study proposes a “Dual-Agent LLM Debate Mechanism” framework using a Proponent (LLM1: Gemini Pro 3) and an Opponent (LLM2: ChatGPT 5.2) to address single-LLM forecasting gaps: The Proponent generates a baseline forecast (F1) from an Integrated Context, while the Opponent validates and resolves conflicts with the Proponent via up to three rounds of cross-debate to produce a consensus forecast (F2). A controlled experiment was conducted to analyze 75 financial market indicators (FMIs) across five asset categories, revealing that F2 outperforms F1 in accuracy and directional stability, particularly in highly volatile assets like Cryptocurrencies and 10-Year Government Bonds. Paired-sample t-tests confirmed statistical significance, validating the mechanism’s effectiveness. Our study results demonstrate how cross-debate between LLMs enhances forecasting accuracy through structured optimization. Full article

Surface electromyography (sEMG) is widely used for gesture recognition, yet the way classic feature–classifier pipelines fail under realistic signal degradations is still poorly quantified. Existing studies typically report accuracy on clean laboratory data, leaving open how amplitude saturation and channel dropout jointly affect different feature combinations, classifiers, and subjects. In this work, we provide, to our knowledge, the first systematic robustness map of a conventional sEMG pipeline under controlledclipping and single-sensor failure. sEMG from nine subjects performing a multi-session, multi-gesture protocol is windowed (250 ms, 50 ms hop) and represented using four common time-domain features (Root Mean Square, Variance, Zero Crossing, and Waveform Length). We exhaustively evaluated single features and all pairwise fusions with three standard classifiers (Support Vector Machine (RBF kernel), Linear Discriminant Analysis, and Random Forest) over (i) a sweep of symmetric saturation thresholds (${10}^{-6}$–${10}^{-1}$) and (ii) five single-channel dropout scenarios, reporting subject-wise dispersion rather than aggregate scores alone. This design enables explicit characterization of the following: (1) accuracy recovery as clipping weakens for each feature pair; (2) dependency of robustness on which channel fails; and (3) differences among Support Vector Machine, Linear Discriminant Analysis, and Random Forest under identical degradations. The results show that lightweight feature pairs (Root Mean Square + Waveform Length, Variance + Zero Crossing, and Waveform Length + Zero Crossing) coupled with Random Forest form a consistently robust operating point, with performance recovering as clipping weakens and remaining resilient under single-channel dropout. Beyond robustness, the conventional pipeline trains substantially faster than representative deep learning baselines under a unified end-to-end timing definition, supporting real-time recalibration and repeated robustness sweeps in wearable deployments. Full article

Reliable identification of seafood species is critical for fisheries management and product authentication, especially when morphological characteristics are lost during processing. In this study, a multiplex PCR system was developed to distinguish seven cuttlefish species (six Sepia spp. and Sepiella inermis) commercially distributed in the Korean seafood market. Species identity was first confirmed by amplifying a mitochondrial cytochrome c oxidase subunit I (COI) fragment (~658 bp) using universal primers (LCO1490/HCO2198), showing 99–100% sequence similarity to corresponding GenBank reference sequences. Analysis of genetic variation based on a 530 bp aligned region demonstrated complete interspecific differentiation without shared haplotypes among species. The number of haplotypes per species ranged from 5 to 21, with haplotype diversity values between 0.667 and 1.000. An extended COI fragment (~1200 bp) was further analyzed to identify diagnostic interspecific variation for marker development. Seven diagnostic single-nucleotide polymorphism (SNP) sites were identified and used to design species-specific forward primers with diagnostic nucleotides positioned at the 3′ termini. Distinct amplicons (220–1099 bp) were generated and clearly resolved by agarose gel electrophoresis. Because simultaneous amplification of all seven primer pairs reduced amplification efficiency, the assay was divided into two multiplex sets. Under optimized conditions (56 °C), each species produced a single expected band without cross-amplification. This multiplex PCR system provides a rapid and sequencing-free approach for reliable species discrimination and can be effectively applied to fisheries monitoring and seafood authentication in commercial supply chains. Full article

Background/Objectives: Delayed graft function (DGF) remains a common early complication after deceased donor kidney transplantation and is challenging to anticipate using routine pre-implant clinical variables alone. We investigated whether high-throughput Fourier transform infrared (FTIR) spectroscopy of static cold storage preservation fluid (not machine perfusion perfusate) captures biochemical information associated with DGF and warrants further evaluation alongside routine pre-implant clinical predictors. Methods: In this single-center retrospective cohort, we analyzed preservation fluid samples from 56 kidney transplants originating from 49 deceased donors (7 donors contributed two kidneys); DGF occurred in 14/56 (25.0%). Dried-film FTIR spectra were acquired using a plate-based high-throughput accessory, and analyses focused on the fingerprint region (900–1800 cm⁻¹) with prespecified preprocessing and quality control. We developed and compared clinical-only, FTIR-only, and combined predictive models and estimated performance using donor-blinded 5-fold StratifiedGroupKFold cross-validation (grouped by donor code) to prevent leakage across paired kidneys. Results: Donor-blinded discrimination (pooled out-of-fold ROC-AUC) was 0.775 for the clinical-only model, 0.814 for the FTIR-only model, and 0.796 for the combined model; probabilistic accuracy (Brier score; lower is better) was 0.162, 0.194, and 0.177, respectively. Calibration intercepts were negative and slopes were <1, indicating overly extreme risk estimates under strict donor-blinded validation and supporting recalibration prior to deployment. Decision curve analysis suggested a positive net benefit for clinically plausible thresholds. Conclusions: These findings support the feasibility of rapid, low-cost FTIR profiling of routinely available preservation fluid as a proof-of-concept approach for exploratory DGF risk stratification, rather than as a clinically deployable prediction tool. Given the small sample size and the instability of subgroup estimates, the main next steps are external validation in larger multicenter cohorts, prospective workflow studies, and model updating/recalibration. Full article

Image classification in low-data regimes remains a challenging problem, particularly in stylized visual domains where intra-class similarity and inter-class feature overlap limit discriminative capacity. This study presents a systematic evaluation of regularization and transfer learning strategies for multi-class comic character recognition under constrained data conditions. Four convolutional architectures are compared: (i) a baseline CNN trained from scratch, (ii) a regularized CNN incorporating data augmentation, dropout, and early stopping, (iii) a pretrained ResNet-50 used as a fixed feature extractor, and (iv) a partially fine-tuned ResNet-50 with selective layer unfreezing. Experiments are conducted on a custom four-class dataset exhibiting moderate class imbalance, evaluated using both a fixed 70/20/10 split and 5-fold cross-validation to assess generalization stability. Results indicate that shallow CNN architectures suffer from substantial overfitting, even when regularization is applied, whereas transfer learning significantly improves macro-averaged F1-score and out-of-distribution detection performance. Cross-validated results, the primary basis for inference given the dataset scale, show that both ResNet-50 strategies achieve equivalent mean accuracy of 95.0% (SD: ±0.4% for feature extraction, ±0.8% for fine-tuning; paired t = 0.00, p = 1.000), while shallow CNN architectures reach only 81–87%. Under a single fixed 70/20/10 partition (n = 69 test samples, 95% CI: ±9–12%), fine-tuning nominally reaches 98.5%; crucially, cross-validation deflates this figure to parity with feature extraction, confirming it reflects favorable partitioning rather than genuine architectural superiority. The primary finding is therefore that frozen ResNet-50 feature extraction is the recommended strategy: it matches fine-tuning in cross-validated generalization while requiring 15× fewer trainable parameters and exhibiting lower fold-to-fold variance. The findings demonstrate that pretrained deep residual representations transfer effectively to stylized comic imagery and that evaluation protocol selection critically impacts perceived performance in small datasets. These results provide practical guidelines for robust model selection in domain-specific, limited-data image classification tasks. Full article

Image restoration under adverse weather conditions has attracted increasing attention because of its importance for both human perception and downstream vision applications. Existing methods, however, are often designed for a single degradation type. We present WeatherMAR, a multi-weather restoration framework that formulates adverse-weather restoration as a paired-domain completion problem in a shared continuous token space. Specifically, WeatherMAR concatenates degraded and clean token sequences into a joint paired-domain sequence and performs restoration through masked autoregressive modeling, in which self-attention enables direct cross-domain interaction. To strengthen conditional learning while avoiding trivial paired correspondences, we introduce complementary bidirectional masking together with an optional reverse objective used only during training to encourage degradation-aware representations. WeatherMAR further employs a conditional diffusion objective for continuous token prediction and adopts a progress-to-step schedule to improve inference efficiency. Extensive experiments on standard multi-weather benchmarks, including Snow100K, Outdoor-Rain, and RainDrop, show that WeatherMAR achieves the best PSNR/SSIM on Snow100K-S (38.14/0.9684), the best SSIM on Outdoor-Rain (0.9396), and the best PSNR on Snow100K-L (32.58) and RainDrop (33.12). These results demonstrate that paired-domain token completion provides an effective solution for adverse-weather restoration. Full article

Growing amounts of municipal waste and the need for efficient recycling demand automated and accurate classification systems. This paper investigates deep learning approaches for multi-class waste sorting based on image data, comparing three widely used convolutional neural network architectures (ResNet-50, EfficientNet-B0, and MobileNet V3) with a custom hybrid model (CustomNet). The dataset comprises 13,933 RGB images across 10 waste categories, combining publicly available samples from the Kaggle Garbage Classification dataset (61.1%) with images collected in house (38.9%). The three glass sub-categories (brown, green, and white glass) were merged into a single glass class to ensure consistent class representation across all dataset splits. Preprocessing steps include normalization, resizing, and extensive data augmentation to improve robustness and mitigate class imbalance. Transfer learning is applied to pretrained models, while CustomNet integrates feature representations from multiple backbones using projection layers and attention mechanisms. Performance is evaluated using accuracy, macro-F1, and ROC–AUC on a held-out test set. Statistical significance was assessed using paired t-tests and Wilcoxon signed-rank tests with Bonferroni correction across five-fold cross-validation runs. The results show that CustomNet achieves 97.79% accuracy, a macro-F1 score of 0.973, and a ROC–AUC of 0.992. CustomNet significantly outperforms EfficientNet-B0 and MobileNet V3 ($p<0.001$, Bonferroni corrected), and it achieves performance parity with ResNet-50 ($p=0.383$) at a substantially lower parameter count in the classification head (9.7 M vs. 25.6 M). These findings indicate that combining multiple feature extractors with attention mechanisms improves classification performance, supports qualitative model explainability via saliency visualization (Grad-CAM), and enables practical deployment on heterogeneous Edge AI platforms. Inference benchmarking on an NVIDIA Jetson Orin Nano demonstrated real-world deployment feasibility at 86.70 ms per image (11.5 FPS). Full article

To meet the requirements of quantitative elemental analysis in the ultraviolet (UV) spectrum, a UV single-photon imaging system was developed, integrating a digital micromirror device (DMD) and a single photon-counting imaging detector, enabling high sensitivity, high resolution, and a wide dynamic range. However, intrinsic geometric distortion poses a significant challenge to accurate spectral calibration. A hybrid correction framework is proposed, cascading polynomial coarse correction with multilayer perceptron (MLP) fine regression, improving calibration accuracy. The method utilizes a full-field dot-array mask projected by the DMD to acquire distortion-reference image pairs. The polynomial model rapidly captures the dominant high-order distortion, while a lightweight MLP performs non-parametric fine regression of residual displacements, achieving a mean error of 0.84 pixels. This approach reduces the root mean square (RMS) error to 1.01 pixels, outperforming traditional direct linear transformation (5.35 pixels) and pure polynomial models (1.33 pixels), while the nonlinearity index decreases from 0.35° to 0.05°. In addition, the method demonstrates stable performance across multi-scale checkerboard patterns ranging from 128 to 280 pixels, with RMS errors remaining around the 1-pixel level. These results validate the high-precision distortion suppression and robust cross-scale performance of the proposed framework. By leveraging DMD-generated patterns for self-calibration, this method eliminates the need for external targets, offering a scalable solution for high-end spectrometer calibration. Full article

Objectives: Clinical trials are essential for therapeutic innovation in rheumatology. A recent decline in clinical trial activity in Hungary has highlighted the need to better understand patient experiences and motivations. This study assessed patient satisfaction and motivation in clinical trials, compared these with routine specialist care, and evaluated healthcare professionals’ motivations. Methods: In this single-center, cross-sectional study, 129 patients completed self-administered questionnaires (61 trial participants and 68 receiving routine care) primarily using a 6-point Likert scale; additionally, 21 healthcare professionals rated their motivations on a Visual Analog Scale (VAS 0–10). Categorical variables were analyzed using chi-square or Fisher’s exact tests, and continuous variables using paired two-tailed t-tests. Results: The main drivers of trial participation were physician recommendations (100%) and trust in the treating physician (100%). Access to novel therapies (98%), closer monitoring (83%), and additional diagnostic procedures (95%) were also significant motivators. Trial participants reported significantly higher satisfaction compared with routine care in terms of consultation time (97% vs. 36%, p < 0.001), staff availability (95% vs. 41, p < 0.001), assistance (93% vs. 36%, p < 0.001), and visit organization (98% vs. 34%; p < 0.001). Overall satisfaction with routine care remained high in both groups. In the control group, fears of disease worsening and the burden of frequent visits were key deterrents. Among healthcare professionals, access to innovative treatments was the strongest motivator, while administrative workload and documentation demands were the primary barriers. Conclusions: Clinical trial participation is associated with high patient satisfaction, driven by physician–patient trust and structured, personalized care. Reducing administrative burdens may be crucial for sustaining clinical research in academic settings. Full article

Cross-institutional collaborative fraud detection is essential for combating increasingly sophisticated financial fraud, yet privacy regulations and data silos severely constrain knowledge sharing among institutions. This study aims to develop a privacy-preserving framework that enables effective collaborative fraud detection while protecting raw data, with particular emphasis on exploiting symmetry properties in federated architectures and graph topology analysis. We propose an Adaptive Federated Graph Attention Network (FedGAT), which employs spatio-temporal graph attention mechanisms to capture topological structures and dynamic fraud patterns within institutional transaction networks. The framework introduces a symmetric similarity matrix derived from graph topological features, where the symmetry property ($s_{ij}=s_{ji}$) ensures consistent and unbiased measurement of structural relationships between any pair of institutions. Based on this symmetric similarity metric, an adaptive weighted aggregation mechanism is designed for cross-institutional parameter fusion, enabling balanced knowledge transfer that respects the symmetric collaborative relationship among participating institutions. The symmetric information exchange protocol between local institutions and the central server further guarantees equitable contribution and benefit distribution throughout the federated learning process. The framework is evaluated on the Elliptic Bitcoin transaction dataset and the IEEE-CIS fraud detection dataset, with recall rate and false positive rate as primary performance metrics. Results show that FedGAT achieves a recall of 0.85 and a false-positive rate of 0.038 in single-institution detection, representing approximately 40% and 70% improvements over existing methods, respectively. In collaborative detection across five virtual institutions, the symmetry-aware adaptive aggregation mechanism enables all participants to achieve performance gains exceeding 15% while completely eliminating negative transfer effects observed in simple averaging approaches. This work contributes a novel symmetry-based federated learning framework that balances privacy protection with detection performance, advancing the literature on cross-institutional financial risk management. Full article

This study benchmarks renewable energy source (RES) utilization in the Visegrad Four (V4) against the EU average using Eurostat data for 2014–2022. A multi-layer framework was used to combine technology-specific per-capita indicators, sectoral RES shares, cluster analysis, and panel regression with fixed effects. The EU substantially exceeds V4 in hydropower (774.06 vs. 270.19 kWh/person), wind (972.06 vs. 161.30 kWh/person), and solar technologies. The electricity-sector gap is most pronounced (EU 41.17% vs. V4 18.69%). Paired t-tests confirmed a statistically significant persistent gap (t(8) = −20.78; p < 0.001), consistent with delayed convergence. Cluster analysis assigned all V4 countries to a single moderate-RES tier, structurally separated from Western and Nordic clusters; panel regression confirmed that the V4 coefficient was robustly negative (β = −5.783 to −9.088 pp) even after policy controls, with fossil lock-in (β = −2.404 pp) emerging as the most consistent structural determinant, whereas V4 × fossil lock-in interaction was positive (β = +2.558 pp), suggesting partial mitigation through differentiated pathways. Intra-V4 heterogeneity—Slovakia’s hydropower lock-in, Hungary’s wind prohibition, Poland’s coal dependency, and Czech Republic’s curtailed feed-in tariff—argues against homogeneous policy responses; results support technology-specific strategies (wind/solar PV in Poland/Czech Republic; solar thermal/heat pumps in Hungary/Slovakia) and grid modernisation as cross-cutting priority. Full article