AI

In this study, we propose a homotopy-inspired prompt obfuscation framework to enhance understanding of security and safety vulnerabilities in Large Language Models (LLMs). By systematically applying carefully engineered prompts, we demonstrate how latent model behaviors can be influenced in unexpected ways. Our experiments encompassed 15,732 prompts, including 10,000 high-priority cases, across LLama, Deepseek, KIMI for code generation, and Claude to verify. The results reveal critical insights into current LLM safeguards, highlighting the need for more robust defense mechanisms, reliable detection strategies, and improved resilience. Importantly, this work provides a principled framework for analyzing and mitigating potential weaknesses, with the goal of advancing safe, responsible, and trustworthy AI technologies. Full article

We address a gap in Machine Learning–human alignment research by proposing that methods from Explainable AI (XAI) can be repurposed to quantitatively model human learning. To achieve alignment between human experts and Machine Learning (ML) models, we must first be able to explain the problem-solving strategies of human experts with the same rigor we apply to ML models. To demonstrate this approach, we model expertise in the complex domain of particle accelerator operations. Analyzing 14 years of operational text logs, we construct weighted graphs where nodes represent operational subtasks and edges capture their strategic relationships. We then examine these strategic models across four granularity levels. Our analysis reveals statistically significant changes with expertise at three of four graph levels. Remarkably, despite numerous possible ways to partition subtasks, operators across all expertise levels demonstrate a striking consistency in high-level strategy, partitioning the task into the same three functional communities. This suggests a shared “divide and conquer” cognitive framework. Expertise develops within this stable framework, as experts exhibit greater cognitive flexibility (forming more cross-community connections) and build more refined internal models. The primary contribution of this work is a methodology for creating a quantitative, interpretable baseline of expert human performance. This provides a “ground truth” for future research in alignment between humans and ML models, enabling a new approach to verification: the ML model’s representation of the task can be quantitatively compared against the human expert benchmark to measure their alignment. This paves the way for building safer, more interpretable partnerships between humans and ML models. Full article

Edge computing (EC) has emerged as a paradigm to support computation-intensive Internet of Things (IoT) applications by enabling task offloading to nearby servers. Despite its potential, the inherent heterogeneity of edge resources and the dynamic, unpredictable nature of task arrivals present significant challenges for designing and evaluating effective load balancing strategies. Traditional evaluation methods are limited as follows: physical testbeds lack scalability and flexibility, while abstract simulators often oversimplify network behavior, failing to capture realistic system dynamics. To address these limitations, we present an emulated dynamic edge computing framework (EDECF) designed for evaluating load balancing schemes in EC networks. First, we developed dedicated service models for each EC node within the EDECF and implemented them using the common open research emulator (CORE) platform, thereby providing a scalable, flexible, and realistic environment for testing optimization strategies. Second, we introduced a robust fitness function that explicitly models latency, queue stability, and fairness for metaheuristic-based load balancing under dynamic edge conditions. To assess its effectiveness, this function was incorporated and tested using the following methods: the particle swarm optimization, genetic algorithm, differential evolution and simulated annealing-based load balancing algorithms. In addition, baseline methods such as the round robin and shortest queue techniques were also deployed to demonstrate the framework’s capacity to facilitate rigorous analysis in heterogeneous and time-varying scenarios. Overall, results are presented to demonstrate EDECF’s capability to emulate realistic workloads, capture resource variability at the edge, and support comprehensive evaluation of algorithmic performance across diverse network settings. Thus, this work aims to establish a practical and extensible foundation for researchers and practitioners to design, test, and optimize load balancing strategies in EC environments. Full article

The RAG agentic architecture has demonstrated its ability to transform large language models (LLMs) into agents capable of planning, reasoning, and executing subtasks using external tools or APIs. In the financial sector, one of the main priorities when implementing new technologies—especially in systems like chatbots—is the protection of customer data and the need to maintain customer trust, making the challenges significant. This research presents a robust banking chatbot system that integrates RAG agentic architecture with specialized financial components, setting a new standard in the digital banking sector by prioritizing security, transparency, and functionality. The contributions of this work include the implementation of RAG agentic reasoning and self-correction financial components, and, primarily, the empirical study of the impact of a semantic firewall with online learning in financial RAG agentic systems, evaluated using public benchmarks and standard ranking metrics. Full article

Background: Bone Mineral Density (BMD) plays a crucial role in diagnosing osteoporosis, and early detection is essential to preventing complications such as osteoporotic fractures. However, access to dual-energy X-ray absorptiometry (DXA) screening remains limited in many healthcare settings. Objective: This study presents a fully automated artificial intelligence pipeline for BMD prediction from lumbar spine radiographs to enable opportunistic osteoporosis screening. Methods: The proposed system integrates automatic vertebral segmentation and a machine learning-based regression model for BMD prediction. A YOLO-based instance segmentation model was trained to automatically segment four lumbar vertebrae, achieving a high Intersection over Union (IoU) of 0.9. Radiomic features were extracted from the segmented vertebrae to capture advanced image characteristics and combined with clinical features from 2875 female patients. An eXtreme Gradient Boosting (XGBoost) regressor was trained to provide opportunistic BMD estimation. Results: The model achieved a mean absolute percentage error (MAPE) of 6% for BMD prediction. A classification model built from segmented vertebrae distinguished between osteoporosis, osteopenia, and normal bone with approximately 90% accuracy. Strong agreement between predicted and ground-truth BMD values was confirmed using Pearson correlation coefficient and Bland–Altman analysis. Conclusions: The proposed fully automated system demonstrates strong agreement with DXA measurements and potential for opportunistic osteoporosis screening in settings with limited DXA access. Further validation and refinement are needed to achieve clinical-grade precision for diagnostic applications. Full article

Safe and reliable railway operation forms an integral part of autonomous transport systems and depends on accurate knowledge of the adhesion conditions. Both the underestimation and overestimation of adhesion can compromise real-time decision-making in traction and braking control, leading to accidents or excessive wear at the wheel–rail interface. Although limited research has explored the estimation of adhesion forces using data-driven algorithms, most existing approaches lack self-reliance and fail to adequately capture low adhesion levels, which are critical to identify. Moreover, obtaining labelled experimental data remains a significant challenge in adopting data-driven solutions for domain-specific problems. This study implements self-reliant deep learning (DL) models as perception modules for intelligent railway systems, enabling low adhesion identification by training on raw time sequences. In the second phase, to address the challenge of label acquisition, a semi-supervised generative adversarial network (SGAN) is developed. Compared to the supervised algorithms, the SGAN achieved superior performance, with 98.38% accuracy, 98.42% precision, and 98.28% F1-score in identifying seven different adhesion conditions. In contrast, the MLP and 1D-CNN models achieved accuracy of 91% and 93.88%, respectively. These findings demonstrate the potential of SGAN-based data-driven perception for enhancing autonomy, adaptability, and fault diagnosis in intelligent rail and robotic mobility systems. The proposed approach offers an efficient and scalable solution for real-time railway condition monitoring and fault identification, eliminating the overhead associated with manual data labelling. Full article

Background: Large language models (LLMs) have demonstrated strong performance on general medical examinations. Whether this performance translates to highly specialized, subspecialty-level board examinations remains unclear. This study evaluates the accuracy and inter-run stability of contemporary LLMs using authentic European Board of Nuclear Medicine (EBNM) Fellowship Examination material. Methods: Ten LLMs (five proprietary, five open-source) completed 50 EBNM multiple-choice questions across five independent zero-shot runs, resulting in 2500 total inferences. Accuracy was calculated per model across runs. Inter-run reliability was assessed using pairwise Cohen’s kappa coefficients. Pairwise model differences were analyzed using McNemar’s test with Bonferroni correction (α = 0.0011). Results: Mean accuracy ranged from 53.6% to 100.0%, with all models exceeding an illustrative 50% pass threshold. Inter-run reliability varied substantially (κ = 0.370–1.000; mean κ = 0.716). High accuracy did not consistently correspond to high reproducibility. Gemini 2.5 Pro achieved high accuracy (93.6%) but showed the lowest reliability (κ = 0.370), whereas DeepSeek V3.2 demonstrated perfect accuracy and agreement across all runs. No significant correlation between accuracy and reliability was observed (Spearman ρ = 0.394, p = 0.26). Conclusions: LLMs demonstrate strong but heterogeneous performance on high-stakes medical knowledge assessments. Differences in reproducibility highlight the need for multi-run evaluation when considering LLMs for educational or clinical knowledge-support applications and for continued validation using non-disclosed examination material. Full article

Ultra-dense networks (UDNs) enable next-generation wireless systems by providing high capacity through aggressive base-station densification. However, dense deployments increase interference and energy consumption, making Quality-of-Service (QoS) aware performance evaluation and optimization challenging. Stochastic geometry (SG) provides a tractable framework for modeling large-scale UDNs, but its use is often limited by simplifying assumptions and simulation requirements. In parallel, Deep Learning (DL) offers scalable tools for capturing complex network behavior from data. This paper proposes a scalable analytical and data-driven framework for performance evaluation and energy efficiency (EE) optimization in UDNs. SG-based analysis is used to derive expressions for key metrics, including coverage probability and EE, under practical QoS constraints such as base-station density, transmit power, activation probability, and SINR thresholds. These results are used to construct a supervised learning dataset, where network parameters and SG derived metrics serve as inputs, and simulation outcomes act as labels. A DL model is trained to capture the nonlinear mapping between network configurations and performance metrics. Results show that the proposed framework predicts coverage probability and EE accurately for unseen UDN scenarios while substantially reducing computational complexity compared to conventional SG-based methods, without violating QoS constraints. Full article

Many fields, including psychology, neuroscience, linguistics, computational modeling, and even philosophy, have been investigating the neuroscience of language for many years. Even so, a lack of comprehensive, interdisciplinary guidelines remains for research projects that aim to decode or model language from brain activity. Electroencephalography (EEG) is unique among neuroimaging methods in that it is a non-invasive technique. This review provides a comprehensive examination of the fundamental elements of imagined speech decoding using EEG, offering a tour of the most recent developments and perspectives in linguistic, neurological, and computational approaches over the past decade. It highlights essential findings such as the consistent involvement of sensory–motor brain regions, the strong influence of language abstraction and selection, and the superior classification performance attained with spectral and temporal features. This study was conducted and reported in accordance with the PRISMA 2020 guidelines for systematic reviews. Full article

This paper presents an intelligent UAV-based parking occupancy monitoring system using a lightweight DJI Mini 3 UAV platform and the YOLOv11 object-detection model. A proprietary aerial dataset was collected from a university parking lot and augmented to address data scarcity, defining two task-oriented classes: vehicle and parking. The proposed framework integrates UAV data acquisition, annotation, data augmentation, training, real-time inference, and occupancy computation into a deployable end-to-end pipeline. Experimental results demonstrate strong detection performance and stable real-time inference, achieving competitive precision, recall, and mAP (mean Average Precision) metrics while maintaining high frame rates suitable for real-time deployment. Comparative evaluation against YOLOv8 and YOLOv9 highlights deployment-oriented advantages rather than architectural novelty. The study confirms that UAV-based vision systems can provide a scalable, low-infrastructure solution for real-time parking monitoring and urban mobility applications, contributing an applied, system-level framework focused on integration and deployment feasibility. Full article

Assessments of extreme psychological constructs often face a persistent challenge: the ceiling effect, in which a significant proportion of respondents select the highest score on a scale, thus obscuring meaningful variation within the population. This effect may have profound consequences in studies of extreme psychological constructs. To address this limitation, we present a novel framework that integrates Multistage Testing (MST) with open-ended questions that are automatically analyzed by large language models (LLMs). This hybrid approach adapts the survey questions to the respondent while leveraging LLMs to efficiently and reliably interpret free-text answers from large-scale online surveys. Using a case study on aversion toward cockroaches, we show how our method can effectively eliminate extreme ceiling effects, revealing hidden data distributions that are often obscured by extreme responses to conventional Likert-type survey questions. We also validate our method by comparing LLM performance to expert human annotations. This demonstrates the consistency and reliability of LLMs in evaluating free-text answers. This framework offers a generalizable methodology that enables more precise and sensitive quantitative measurement of extreme psychological constructs, allowing researchers to study topics that until now were inaccessible due to significant, inherent ceiling effects. Full article

Hand localization in cluttered industrial environments remains challenging due to variations in appearance and the gap between synthetic and real-world data. Domain randomization addresses this “reality gap” by intentionally introducing randomized and unrealistic visual features in simulated scenes, encouraging neural networks to focus on essential domain-invariant cues. In this study, we applied domain randomization to generate a synthetic Red-Green-Blue–Depth (RGB-D) dataset for training multimodal instance segmentation models, with the aim of achieving color-agnostic hand localization in complex industrial settings. We introduce a new synthetic dataset tailored to various hand detection tasks and provide ready-to-use pretrained instance segmentation models. To enhance robustness in unstructured environments, the proposed approach employs multimodal inputs that combine color and depth information. To evaluate the contribution of each modality, we analyzed the individual and combined effects of color and depth on model performance. All evaluated models were trained exclusively on the proposed synthetic dataset. Despite the absence of real-world training data, the results demonstrate that our models outperform corresponding models trained on existing state-of-the-art datasets, achieving higher Average Precision and Probability-Based Detection Quality. Full article

Recent advances in computational chemistry, machine learning, and large-scale virtual screening have rapidly expanded the accessible chemical space, increasing the need for interpretable molecular representations that capture the hierarchical topological structure of molecules. Existing formats, such as Simplified Molecular Input Line Entry System (SMILES) strings and MOL files, effectively encode molecular graphs but provide limited support for representing the multi-level structural information needed for complex downstream tasks. To address these challenges, we introduce HiFrAMes, a novel graph–theoretic hierarchical molecular fragmentation framework that decomposes molecular graphs into chemically meaningful substructures and organizes them into hierarchical scaffold representations. HiFrAMes is implemented as a four-stage pipeline consisting of leaf and ring chain extraction, ring mesh reduction, ring enumeration, and linker detection, which iteratively transforms raw molecular graphs into interpretable abstract objects. The framework decomposes molecules into chains, rings, linkers, and scaffolds while retaining global topological relationships. We apply HiFrAMes to both complex and drug-like molecules to generate molecular fragments and scaffold representations that capture structural motifs at multiple levels of abstraction. The resulting fragments are evaluated using selection criteria established in the fragment-based drug discovery literature and qualitative case studies to demonstrate their suitability for downstream computational tasks. Full article

Reservoir computing (RC) has emerged as an energy-efficient paradigm for temporal information processing, offering reduced training complexity by fixing recurrent dynamics and training only a simple readout layer. Among RC models, Echo State Networks (ESNs) and Liquid State Machines (LSMs) represent two distinct approaches based on continuous-valued and spiking neural dynamics, respectively. In this work, we present a comparative evaluation of ESNs and LSMs on the Mackey–Glass chaotic time-series prediction task, with emphasis on scalability, overfitting behavior, and robustness to reduced numerical error precision. Experimental results show that ESNs achieve lower prediction error with relatively small reservoirs but exhibit early performance saturation and signs of overfitting as reservoir size increases. In contrast, LSMs demonstrate more consistent generalization with increasing reservoir size and maintain stable performance under aggressive reservoir quantization. These findings highlight fundamental trade-offs between accuracy and hardware efficiency, and suggest that spiking RC models are well suited for energy-constrained and neuromorphic computing applications. Full article

This study examined whether large language models (LLMs) can generate clinically realistic profiles of cognitive decline and whether simulated deficits reflect architectural constraints rather than superficial role-playing artifacts. Using GPT-4o-mini, we generated synthetic cohorts (n = 10 per group) representing healthy aging, mild cognitive impairment (MCI), and Alzheimer’s disease (AD), assessed through a conversational neuropsychological battery covering episodic memory, verbal fluency, narrative production, orientation, naming, and comprehension. Experiment 1 tested whether synthetic subjects exhibited graded cognitive profiles consistent with clinical progression (Control > MCI > AD). Experiment 2 systematically manipulated prompt context in AD subjects (short, rich biographical, and few-shot prompts) to dissociate robust from manipulable deficits. Significant cognitive gradients emerged (p < 0.001) across eight of thirteen domains. AD subjects showed impaired episodic memory (Cohen’s d = 4.71), increased memory intrusions, and reduced narrative length (d = 3.07). Critically, structurally constrained memory tasks (episodic recall, digit span) were invariant to prompting (p > 0.05), whereas generative tasks (narrative length, verbal fluency) showed high sensitivity (F > 100, p < 0.001). Rich biographical prompts paradoxically increased memory intrusions by 343%, indicating semantic interference rather than cognitive rescue. These results demonstrate that LLMs can serve as in silico test benches for exploring candidate digital biomarkers and clinical training protocols, while highlighting architectural constraints that may inform computational hypotheses about memory and language processing. Full article

Deepfake technology is causing unprecedented threats to the authenticity of digital media, and demand is high for reliable digital media detection systems. This systematic review focuses on an analysis of deepfake detection methods using deep learning approaches, machine learning methods, and the classical methods of image processing from 2018 to 2025 with a specific focus on the trade-off between accuracy, computing efficiency, and cross-dataset generalization. Through lavish analysis of a robust peer-reviewed studies using three benchmark data sets (FaceForensics++, DFDC, Celeb-DF) we expose important truths to bring some of the field’s prevailing assumptions into question. Our analysis produces three important results that radically change the understanding of detection abilities and limitations. Transformer-based architectures have significantly better cross-dataset generalization (11.33% performance decline) than CNN-based (more than 15% decline), at the expense of computation (3–5× more). To the contrary, there is no strong reason to assume the superiority of deep learning, and the performance of traditional machine learning methods (in our case, Random Forest) is quite comparable (accuracy of 99.64% on the DFDC) with dramatically lower computing needs, which opens up the prospects for their application in resource-constrained deployment scenarios. Most critically, we demonstrate deterioration of performance (10–15% on average) systematically across all methodological classes and we provide empirical support for the fact that current detection systems are, to a high degree, learning dataset specific compression artifacts, rather than deepfake characteristics that are generalizable. These results highlight the importance of moving from an accuracy-focused evaluation approach toward more comprehensive evaluation approaches that balance either generalization capability, computational feasibility, or practical deployment constraints, and therefore further direct future research efforts towards designing systems for detection that could be deployed in practical applications. Full article

Radon (Rn-222) is a major indoor air pollutant with significant health risks. This work presents RadonFAN, a low-cost IoT system deployed in two galleries at the Institute of Physical and Information Technologies (ITEFI-CSIC, Madrid), integrating distributed sensors, microcontrollers, cloud analytics, and automated fan control to maintain radon concentrations below recommended limits. Initially, ventilation relied on a reactive, rule-based mechanism triggered when thresholds were exceeded. To improve preventive control, two end-to-end deep learning models based on regression-to-classification (R2C) and direct classification (DC) are developed. A quantitative analysis of predictive performance and computational efficiency is reported. While the R2C model is hindered by the inherent behavior of the time series, the DC model achieves high classification performance (recall > 0.975) with low computational cost (<4 million parameters, 7 million FLOPs). Modifications to the DC model are studied to identify potential performance bottlenecks and the most relevant components, showing that most limitations arise from feature richness and time series behavior. When evaluated against the existing rule-based ventilation system, the DC model reduces both unsafe radon exposure events and energy consumption, demonstrating its effectiveness for preventive radon mitigation. Full article

Modern malware presents significant challenges to traditional detection methods, often leveraging fileless techniques, in-memory execution, and process injection to evade antivirus and signature-based systems. To address these challenges, alert-driven memory forensics has emerged as a critical capability for uncovering stealthy, persistent, and zero-day threats. This study presents a two-stage host-based malware detection framework, that integrates memory forensics, explainable machine learning, and ensemble classification, designed as a post-alert asynchronous SOC workflow balancing forensic depth and operational efficiency. Utilizing the MemMal-D2024 dataset—comprising rich memory forensic artifacts from Windows systems infected with malware samples whose creation metadata spans 2006–2021—the system performs malware detection, using features extracted from volatile memory. In the first stage, an Attentive and Interpretable Learning for structured Tabular data (TabNet) model is used for binary classification (benign vs. malware), leveraging its sequential attention mechanism and built-in explainability. In the second stage, a Voting Classifier ensemble, composed of Light Gradient Boosting Machine (LGBM), eXtreme Gradient Boosting (XGB), and Histogram Gradient Boosting (HGB) models, is used to identify the specific malware family (Trojan, Ransomware, Spyware). To reduce memory dump extraction and analysis time without compromising detection performance, only a curated subset of 24 memory features—operationally selected to reduce acquisition/extraction time and validated via redundancy inspection, model explainability (SHAP/TabNet), and training data correlation analysis —was used during training and runtime, identifying the best trade-off between memory analysis and detection accuracy. The pipeline, which is triggered from host-based Wazuh Security Information and Event Management (SIEM) alerts, achieved 99.97% accuracy in binary detection and 70.17% multiclass accuracy, resulting in an overall performance of 87.02%, including both global and local explainability, ensuring operational transparency and forensic interpretability. This approach provides an efficient and interpretable detection solution used in combination with conventional security tools as an extra layer of defense suitable for modern threat landscapes. Full article

Edge artificial intelligence (AI) workloads generate bursty, heterogeneous traffic on Networks-on-Chip (NoCs) under tight energy and latency constraints. Hybrid NoCs that overlay electronic meshes with silicon photonic express links can reduce long-path latency via wavelength-division multiplexing, but thermal drift and intermittent optical availability complicate routing. This study introduces a decentralized, photonic-aware controller based on Deep Reinforcement Learning (DRL) with Proximal Policy Optimization (PPO). The policy uses router-local observables—per-port buffer occupancy with short histories, hop distance, a local injection estimate, and a per-cycle optical validity signal—and applies action masking so chosen outputs are always feasible; the controller is co-designed with the router pipeline to retain single-cycle decisions and a modest memory footprint. Cycle-accurate simulations with synthetic traffic and benchmark-derived traces evaluate mean packet latency, throughput, and energy per delivered bit against deterministic, adaptive, and recent DRL baselines; ablation studies isolate the roles of optical validity cues and locality. The results show consistent improvements in congestion-forming regimes and on long electronic paths bridged by photonic links, with robustness across mesh sizes and wavelength concurrency. Overall, the evidence indicates that photonic-aware PPO provides a practical, thermally robust control plane for hybrid NoCs and a scalable routing solution for AI-centric manycore and edge systems. Full article