Search Results (7,067)

Physiology-based mental workload classification is hard to compare across studies because task design, preprocessing, segmentation, and validation protocols vary widely. Using OpenNeuro ds007262, an open multimodal arithmetic dataset of synchronised 19-channel 10–20 system electroencephalography (EEG), electrocardiography (ECG), and pupillometry data from 18 released participants (16 retained after participant-level quality control for downstream modelling) spanning seven objective difficulty bands plus baseline fixation, we present a reproducible end-to-end machine learning pipeline for graded workload classification. The pipeline standardises participant-level quality control, trial-aligned windowing, modality-specific preprocessing and feature extraction (153 EEG, 18 ECG, and 25 pupillometry features), and supervised evaluation under three validation protocols (within-participant, pooled-stratified, and group-holdout) over eleven models and five class scenarios. Fused representations generally performed best; EEG was the strongest unimodal modality, and classical models outperformed deep models in most feature-based conditions. The best 6 s pipeline reached a balanced accuracy of 0.635; under denser 3 s overlap segmentation, the best pipeline reached 0.718. Mean balanced accuracy across 60 matched cells rose from 0.324 to 0.380, with gains concentrated in within-participant and pooled-stratified evaluation rather than strict unseen-participant transfer. The pipeline provides a transparent benchmark framework for fine-grained physiological workload modelling. Full article

Source code plagiarism is a significant challenge in software engineering and computer science education, affecting academic integrity, intellectual property rights, and software quality assurance. However, Source Code Plagiarism Classification (SCPC) remains difficult because labelled training data are limited, mainly due to the sensitivity of plagiarism cases. This restricts the effective use of machine learning (ML) and deep learning (DL) methods, especially in low-resource settings. This work investigates low-resource SCPC using Pre-trained Code Models (PCMs). We first examine Full Fine-Tuning (FFT), where all model parameters are updated, across multiple public datasets. We then evaluate Parameter-Efficient Fine-Tuning (PEFT), where only a small subset of parameters is trained. Specifically, we apply three adapter-based PEFT methods and compare them with FFT in terms of classification performance, training time, inference time, GPU usage, trainable parameter percentage, and model size. The results show that, when labelled training data are available, fine-tuned PCMs achieve strong SCPC performance and higher $F_1$ scores than the unsupervised open-source plagiarism-detection tools in our evaluation, JPlag and Dolos. Overall, PEFT achieves a performance that is similar, comparable to, or slightly lower than that of FFT, while requiring fewer trainable parameters and lower GPU usage, at the cost of slightly higher inference time. Full article

Action recognition has emerged as a critical area of research within the realm of computer vision, driven by the increasing demand for intelligent human–machine systems capable of understanding and interpreting human behaviors in the real world. The ability to decipher intricate details of human actions holds immense potential to improve system design, predictive modeling, data-informed decision-making, and real-time operational improvements across a wide variety of domains. Some examples of applications range from surveillance and real-time management of public spaces and infrastructure systems, to development of predictive modeling and robotic systems for individualized healthcare interventions, to implementing effective human–computer interaction in both professional and recreational settings. This paper provides a comprehensive survey of the current state of action recognition, focusing specifically on three open-world challenges: the integration of multimodalities, the ethical and social implications of these technologies, and the utilization of feedback mechanisms to enhance model performance. We delve into the evolution of action recognition, from early feature-based approaches to the deep learning revolution, emphasizing how the incorporation of multiple sensory modalities—such as visual, audio, and depth data as well as other cues—has advanced the field. Furthermore, we examine the ethical challenges associated with deploying these technologies in the public domain, particularly regarding privacy, bias, and societal impact, and discuss the need for responsible development and regulation. The third focus of the paper is the use of top-down and bottom-up feedback mechanisms within deep learning architectures, exploring how these strategies can mimic human cognitive processes to improve accuracy and reliability in action recognition systems. By identifying current gaps and proposing future research directions, this paper aims to inspire continued innovation in this dynamic and impactful field for intelligent systems. Full article

Background: Intelligent decision support systems for colonoscopy can fail when encountering out-of-distribution surgical instruments such as snares or biopsy forceps, producing false-positive polyp masks that may contribute to automation bias and reduce workflow reliability. This study aimed to develop and evaluate a parameter-efficient framework for open-set-aware polyp segmentation that can reject such anomalous inputs while preserving in-distribution segmentation performance. Methods: We propose PolypSAM-Open, which integrates a prototype-based Open-Set Learning (OSL) module with Low-Rank Adaptation (LoRA) in the MedSAM image encoder. The model was trained on Kvasir-SEG using an 85/15 split of authentic polyp images and synthetic high-frequency Gaussian noise to learn a rejection margin. Zero-shot out-of-distribution detection was evaluated on 590 unseen authentic surgical instruments from Kvasir-Instrument. Segmentation and detection performance were compared against a standard MedSAM-LoRA baseline. Results: Standard parameter-efficient fine-tuning yielded an OOD AUROC of 0.4263 on authentic surgical instruments. PolypSAM-Open improved zero-shot OOD AUROC to 0.9535 ($p<0.001$). Despite allocating 15% of training capacity to the synthetic-noise rejection margin, PolypSAM-Open maintained segmentation performance comparable to the standard fine-tuned baseline (Dice 0.9728 versus 0.9723). On ETIS-LaribPolypDB and CVC-ClinicDB, Dice scores were 0.9301 and 0.9386, respectively. The approach remained parameter-efficient, updating 4.48% of total parameters while adding negligible inference latency relative to the underlying MedSAM forward. Conclusions: Prototype-based open-set adaptation can substantially improve rejection of unseen surgical artifacts in AI-assisted colonoscopy while preserving high segmentation accuracy. These findings position PolypSAM-Open as a promising strategy for potentially safer decision-support segmentation in endoscopic workflows; prospective clinical validation remains necessary. Full article

Cadmium (Cd) is a widespread environmental pollutant associated with neurotoxicity, but its underlying mechanisms remain unclear. Disulfidptosis is a regulated cell death driven by disulfide stress under conditions of impaired cellular reducing capacity. This study investigated the potential involvement of disulfidptosis-associated molecular alterations in Cd-induced neurotoxicity. Male Sprague Dawley (SD) rats were exposed to cadmium chloride (Low-Dose Group: CdCl₂: 0.036 mg/kg bw; High-Dose Group: CdCl₂: 3.6 mg/kg bw) by oral gavage for 30 days. Neurobehavioral performance was assessed using the open field test, elevated plus maze, and Morris water maze. Hippocampal ultrastructure, redox-related metabolites, and disulfidptosis-associated genes were analyzed. In addition, bioinformatics analysis was performed by integrating cadmium-related, neurodegenerative disease-related, and disulfidptosis-related genes. The results showed that high-dose Cd exposure impaired locomotor activity, increased anxiety-like behavior, and disrupted spatial learning and memory (p < 0.05), accompanied by mitochondrial damage in hippocampal neurons. Bioinformatics analysis identified seven overlapping genes and enrichment of ferroptosis and oxidative phosphorylation pathways. Biochemically, cadmium exposure significantly increased the NADP⁺/NADPH ratio ([Control: 1.07 ± 0.044] vs. [High-dose: 3.80 ± 0.059], p < 0.05) and decreased the GSH/GSSG ratio ([Control: 2.80 ± 0.059] vs. [High-dose: 1.14 ± 0.091], p < 0.05), indicating severe redox imbalance. At the molecular level, cadmium exposure upregulated SLC7A11 mRNA expression by 1.48 ± 0.12-fold (p < 0.01) and SLC3A2 by 1.91 ± 0.55-fold (p < 0.05), while downregulating NDUFS1 expression to 0.84 ± 0.01-fold of control levels (p < 0.01) in hippocampal tissues. These findings suggest that high-dose Cd exposure induced neurotoxicity is associated with mitochondrial dysfunction, redox imbalance, and disulfidptosis-associated molecular alterations. Full article

The proliferation of mobile communications has led to a significant increase in SMS spam, posing challenges related to security, privacy, and user experience. Although numerous machine-learning-based spam detection approaches have been proposed, developing systems that are simultaneously lightweight and resilient to adversarial manipulation remains an open problem. This paper proposes an SMS spam detection framework that incorporates multiple feature extraction methods, including bag-of-words (BoW), Term Frequency–Inverse Document Frequency (TF-IDF), and N-gram models with dimensionality reduction using principal component analysis (PCA), followed by classification using decision tree (DT) and Logistic Regression (LogReg) models. Experimental evaluations on the UCI SMS Spam Collection dataset demonstrate that the TF-IDF-PCA-DT pipeline achieves a detection accuracy of 99% while reducing model size by 77% and inference time by 75%. Robustness evaluation under adversarial text perturbations indicates minimal performance degradation, maintaining an accuracy of 96.5%. These findings demonstrate the practicality of the proposed framework for real-world deployment in resource-constrained environments. Full article

Lead-ion (Pb²⁺) contamination in drinking water poses a serious threat to public health, but conventional laboratory-based methods rely on bulky equipment and are unsuitable for on-site monitoring. Here, we develop a wireless monitoring system based on a dual-channel liquid–solid triboelectric nanogenerator probe (TENG probe) for detecting Pb²⁺ in drinking water. Based on the dynamic contact electrification at the liquid–solid interface, the sliding of a water droplet containing Pb²⁺ on the FEP surface is converted into an electrical signal for Pb²⁺ detection. A wireless acquisition circuit transmits the electrical signals via Wi-Fi to a computer, enabling remote and wireless detection. By integrating a one-dimensional convolutional neural network (1D CNN) deep learning model, the TENG probe achieved a detection accuracy of 99.62% and was capable of detecting Pb²⁺ in drinking water at the ppb level, exceeding the national standard. This work opens a way for safeguarding drinking-water quality. Full article

Urban last-mile delivery is a rapidly growing source of city-centre emissions, and decarbonizing it without eroding service quality has become imperative for climate goals. Operators are turning to multimodal systems that integrate road vehicles, off-peak metro freight, and electric drones—yet the optimal delivery channel varies dynamically with location and time. Current RL-based schedulers handle constraints via manually tuned penalty weights, lacking formal safety guarantees, and the feasibility of online carbon-cap enforcement under partial observability remains an open question. To address this, we model the problem as a Constrained Markov Decision Process (CMDP) and propose a demonstration-regularized Lagrangian deep RL algorithm. Our approach learns an online policy that is model-free at deployment—it controls emissions in expectation against a hard carbon budget, makes per-order decisions using only state observations, and operates without an emission model at test time (the demonstrator used at training time does access the emissions model, so “model-free” refers strictly to the deployment phase). Experiments on synthetic benchmarks and a Nanjing-inspired scenario—grounded in real metro topology and population-weighted demand—show that our policy achieves emissions within 1.3% of the offline optimum. It robustly tracks a ±17% carbon-budget band across a threefold daily volume range and a threefold city-scale range, with zero per-instance tuning. By contrast, a standard PPO with fixed penalty weights consistently degrades to single-mode selection. Our findings suggest that hard carbon budgets can be controlled in expectation online at modest cost—a step toward operator-facing low-carbon logistics whose average emissions honour a binding carbon budget, though external validation on operational data and a risk-sensitive formulation that upgrades this average control into per-day compliance are still required before deployment. Full article

Cotton fiber quality is shaped during boll development, boll opening, fluffing, and harvesting, but current assessment still relies largely on manual field inspection and postharvest laboratory testing. This limits timely harvest scheduling and plot-level quality management. To address this problem, we propose a self-supervised multimodal sensing framework for linking preharvest cotton boll status, environmental conditions, and postharvest fiber quality. First, the Cotton Boll Visual Phenotype Self-Supervised Encoding Module learns maturity-related visual representations by reconstructing masked image patches, so that boll cracking, lint exposure, and surface texture can be captured from unlabeled field images. Second, the Agricultural Sensor Temporal Masked Modeling Module reconstructs masked sensor observations to model temporal patterns in temperature, humidity, light, soil moisture, rainfall, and other environmental variables. Third, the Vision–Environment Cross-Modal Contrastive Fusion Module aligns image features with environmental features and produces a joint representation for downstream prediction. Field experiments were conducted using cotton boll images from different maturity and abnormal states, environmental sensor records, management information, and postharvest fiber quality measurements. The framework was evaluated for maturity classification, harvest-window recognition, and fiber quality prediction. The results showed that the proposed method performed consistently better than representative machine learning, single-modal deep learning, and multimodal fusion baselines, while few-shot and ablation experiments supported the value of self-supervised pretraining and multimodal fusion. These findings indicate that the proposed approach can provide useful information for preharvest cotton maturity assessment and harvest-quality management. Full article

As a fundamental and highly challenging task in the field of computer vision, industrial object counting plays a critical role in smart manufacturing, inventory management, and production process monitoring. Over the past fifteen years (2010–2025), this field has undergone a profound technological transformation, shifting from traditional machine vision methods relying on handcrafted features to a data-driven paradigm based on deep learning. This paper aims to provide a comprehensive and systematic review of this rapidly evolving research area, with technological evolution as the core narrative thread. First, we review early traditional methods, analyzing the application of sensor-based and template-matching technologies in controlled environments, as well as their core limitations in complex industrial scenarios. Subsequently, this paper focuses on exploring how the introduction of deep learning has reshaped the landscape of counting tasks, and elaborates on the breakthrough progress of convolutional neural networks (CNNs), Transformer architectures, the recently emerging Mamba state space model, and Large Foundation Models in addressing key challenges including occlusion, object overlap, multi-scale variation, and dense object counting. In particular, this paper conducts an in-depth analysis of the paradigm shift from Class-Specific Counting to Class-Agnostic Counting (CAC) and Exemplar-Free Counting. This trend significantly reduces the reliance on large-scale annotated data and greatly enhances the generalization ability of models in open-world scenarios. Additionally, this paper systematically organizes mainstream datasets in the field, including FSC-147, NWPU-MOC, and OmniCount-191, and compares core evaluation metrics such as Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and the PrACo metric system. In response to the core technical challenges faced by current methods, including high annotation costs, weak cross-domain adaptability, and strict real-time requirements in industrial scenarios, this paper proposes key future research directions including lightweight model design, unsupervised learning, multi-modal fusion, and Prompt-based interactive counting. This review intends to provide researchers in both academia and industry with a complete technical blueprint so as to promote the continuous development of industrial object-counting technology toward a more efficient and intelligent direction. Full article

Additive manufacturing has opened an extraordinary design space for polymer metamaterials, enabling microstructures whose macroscopic mechanical behavior is governed largely by geometry rather than by chemical composition. A principled design framework must solve two coupled problems: a forward problem (given a microstructure, predict effective properties) and an inverse problem (given target properties, generate a microstructure). Convolutional neural networks (CNNs) solve the forward problem accurately, but the inverse problem remains more challenging for three reasons reported in the literature: (i) many surrogates predict only a scalar proxy rather than the full second-order elastic tensor; (ii) fixed or randomly initialized inverse decoders create a distribution-shift gap between surrogate predictions and physical re-evaluation; and (iii) dataset bias toward near-solid configurations limits exploration of low-density and anisotropic designs. We present a unified framework, the Operator-Physics-Enhanced Reverse Architecture (OPERA), that addresses all three issues. First, the forward surrogate predicts the complete $3\times 3$ plane-stress stiffness tensor $C^{\mathrm{eff}}$ in Voigt notation, with an analytical layer enforcing $C_{ij}=C_{ji}$ and positive definiteness by construction, achieving $R^2>0.99$ on the directional moduli and density and $R^2>0.88$ on the off-diagonal coupling term $C_{16}$ and the effective Poisson ratio. Second, a normalizing-flow decoder ${\mathcal{F}}_{\varphi }$, jointly trained with the forward surrogate, keeps inverse design on the training manifold and reduces the surrogate–PDE re-evaluation gap from more than 30% to below 6% on held-out targets. Third, a five-family dataset with uniform coverage of $\rho \in [0.10,0.95]$ is augmented through an expected-improvement active-learning loop. We embed minimum-feature-size, connectivity, and print-direction constraints into the optimization through differentiable regularization and report agreement of $R^2=0.987$ between predictions and tensile measurements on ten FDM-printed specimens. The framework is demonstrated on five problems (auxetic, extreme anisotropy, isotropic low-density, chiral, and hierarchical), with an average target error of 6.8%. The results are framed relative to a reproduced scalar-proxy baseline; we provide an explicit statistical uncertainty analysis, a baseline-reproduction protocol, and a discussion of the method’s assumptions and numerical enforcement. Full article

Patient-specific dosimetry and quantitative external counting for in vivo monitoring of radiopharmaceutical therapies require detector calibration coefficients (k) that are currently obtained through computationally intensive Monte Carlo (MC) simulations. We present a digital-twin framework that learns the mapping from organ-level Specific Absorbed Fractions (SAFs) to clinical k values for X-ray and gamma-ray detectors, trained on 4.47 million SAF entries from the OpenDose collaboration covering the ICRP-110 adult male (AM) and adult female (AF) reference phantoms, 91 photon energies (5–10,000 keV), and 336 source organs. An uncertainty-aware ensemble combining Histogram Gradient Boosting, ExtraTrees, Random Forest, Quantile-HGBT, and a Feature-Tokenizer Transformer (414 K parameters) was stacked via ridge regression and calibrated using Conformalized Quantile Regression. The ensemble achieved a test mean absolute error of 0.220 in log10-SAF space (R² = 0.921) with an empirical 95% prediction interval coverage of 95.0%. Validation against 10,487 independent published S-values (EMDOSE) yielded a Pearson correlation of 0.990 (log space). The framework reduces k-coefficient computation time from hours of MC simulation to milliseconds, supporting real-time detector calibration for theranostic workflows. Full article

This study explores the introduction of investigative approaches to preservice mathematics teachers in countries where traditional exercise-oriented teaching remains dominant despite ongoing educational reforms promoting student-centered, competency-based learning. The Albanian mathematics education is taken as a case study representative of many other countries coming from highly structured teacher-centered mathematics education systems to more progressive models, where this implementation gap persists, and where curriculum changes are not always accompanied by attention to the local situations. The purpose of the study is to examine the characteristics of critical mathematical competence exhibited by preservice teachers when they engage in investigative activities and to identify the possibilities and barriers they perceive. A qualitative case study was conducted at an Albanian university with students of the master’s program in mathematics teacher education. The intervention included pattern investigations, mathematical modelling tasks, and an open-ended investigative task on food security. Data consisted of written reflections, group work recordings, written assignments, and interviews. The results show that participants exhibited mathematical, technological, and, to some extent, reflective knowing. They used mathematics to explore real-world problems about food security, evaluated data sources, and reflected on their validity. At the same time, they faced several barriers, including traditional orientations about mathematics teaching and lack of experience with investigative approaches, uncertainty when working with open-ended tasks, concerns about the reliability of data sources, and time constraints. Participants also identified important possibilities, such as student engagement, collaboration, critical thinking, and stronger connections between mathematics and real-life contexts. The study highlights both the potential and the complexity of integrating investigative approaches into mathematics teacher education in Albania, but also to other countries undergoing similar curricular reforms. Full article

Recently, the yoga industry has witnessed rapid technology integration, particularly through computer vision and machine learning methods, to improve the practice experience. We developed a yoga posture detection and feedback system using a Raspberry Pi 5-based hardware setup and a combination of the Open Source Computer Vision Library, MediaPipe, and TensorFlow. The system records videos of practitioners doing five standard yoga poses and provides rapid data-based feedback to help fix misalignments. The model accurately classifies poses using convolutional neural networks and identifies important joint landmarks and limb angles, which is confirmed through confusion matrix analysis. The developed system lets its users practice safely without direct supervision from an instructor. It increases access to practical yoga training. The setup also allows for future physical therapy and athletic training applications, where proper form and alignment matter. Full article

Frontier large language models achieve broad linguistic competence but degrade on specialist domains underrepresented in pre-training corpora. Domain-adaptive post-training (DAPT) on curated professional text partially closes this gap, yet the dominant approach flattens structured discourse into isolated document units, discarding the collaborative reasoning embedded in multi-party exchanges. This paper investigates whether preserving the full recursive structure of user forum threads during post-training is a more effective first step toward knowledge extraction than flattened question-answer pairs. Four open-source decoder-only models (TinyLlama 1.1B, Phi-2 2.7B, LLaMA-2-7B 6.8B, LLaMA-2-13B 13B) are post-trained using parameter-efficient LoRA adaptation on 4970 threads from AgTalk, an agricultural producer forum, under three conditions: flattened Q → A pairs, full recursive threads preserving reply order, and shuffled recursive threads with randomly permuted intermediate replies. Five hypotheses are tested through paired Wilcoxon signed-rank comparisons across 29 thread positions. DAPT significantly reduces perplexity relative to pretrained baselines across all architectures (H0 supported). Recursive training reduces perplexity relative to flattened training, an advantage clearest for the two LLaMA-2 models under matched-context evaluation (Wilcoxon win rates near 72%) and present but obscured by outlier variance at the 1.1B and 2.7B scales (H₁ supported). However, ordered recursive training provides only a marginal advantage over shuffled (H₂ inconclusive), attention analysis reveals identical U-shaped endpoint-weighted profiles regardless of training condition (H₃: architectural not learned), and perplexity shows no systematic decrease with accumulating thread depth (H₄ not supported). These results are attributed to Rotary Position Embedding constraints in decoder-only architectures that systematically underweight middle-thread content. Encoder–decoder architectures with bidirectional cross-attention are identified as a promising next step for exploiting the full collaborative structure of forum discourse. Full article