Evaluating the Effectiveness of Q-Learning in Ant Colony Optimization for the Traveling Salesman Problem

The Traveling Salesman Problem (TSP) is a fundamental combinatorial optimization challenge where exact solvers face exponential time complexity. Consequently, metaheuristics are widely employed to identify high-quality solutions within feasible timeframes. In this work, we investigate the added utility of integrating Reinforcement Learning into swarm intelligence by proposing QL-ACO, a hybrid framework that naturally embeds Q-learning into the Ant Colony Optimization (ACO) metaheuristic. The primary objective is to assess how Q-learning affects the trade-off between exploration and exploitation by simultaneously updating Q-values and pheromone trails. This hybrid search is further refined using a 2-Opt local search mechanism. We evaluate the proposed approach on standard TSPLIB benchmark instances to analyze both solution optimality and algorithmic efficiency. The results demonstrate an average optimality gap of 2.19%, which is comparable to state-of-the-art hybrid algorithms. Moreover, to verify the specific added value of the reinforcement learning component, we conduct a comprehensive ablation study. This analysis reveals that the inclusion of Q-learning significantly accelerates convergence compared to the baseline ACO (by up to 12%), confirming its effectiveness in guiding the search process toward high-quality solutions more efficiently.

RSCAAT: A Recursive Mechanism to Mitigate Cache Side Channel Attacks

Multi-level caching was introduced to increase computing performance by alleviating the speed difference between the processors and memory. However, this has introduced vulnerabilities that can be exploited by logical side-channel attacks. While there exist solutions that perform run-time remapping of data under threat to randomized locations, such solutions inhibit vulnerabilities in case of repetitive attacks. This paper goes beyond the state of the art by proposing a run-time mitigation mechanism called Recursive Secure Cache Alternative Address Table (RSCAAT) designed with a robust control logic and enhanced reactivity to cache behavior. It includes a specialized Content-Addressable Memory (CAM) with built-in remapping redundancies capable of eliminating predictable pseudorandom traces resulting in area overheads in the range of 4-5%. RSCAAT is evaluated using benchmarks from several suites. Results show that while the overhead of RSCAAT in performance is comparable to the state of the art, it provides significantly higher levels of security to the cache by accurately eliminating leakage-like patterns as well as recursively remapping vulnerable accesses. RSCAAT is activated for 100% of the leakage-like patterns detected while state-of-the-art solutions have been shown to only activate 53-93%.

Robotic Haircutting Systems: A Survey of Methods, Challenges, and Hair Modeling Insights

The field of robotic haircutting is rapidly evolving, merging advances in service robotics, computer vision, force sensing, and deep learning. This survey paper highlights the growing importance of robotic systems in personal grooming, driven by demographic shifts, technological innovation, and increasing demand for hygienic, consistent services. We review the historical evolution of robotic haircutting technologies, including DIY solutions, academic prototypes, and emerging commercial systems. Core technologies such as robotic arms, hair modeling, motion planning, and tactile sensing are discussed in detail. Key challenges including hair variability, real-time decision-making, user customization, and safety considerations are analyzed. Furthermore, we explore the transformative role of virtual hairstyle modeling and the integration of multimodal perception and AI-driven personalization. Future research directions emphasize the incorporation of large language models, vision-language models, reinforcement learning, and diffusion models to advance human-robot collaboration in salons, while addressing ethical, societal, and inclusivity considerations.

A Novel Multi-Scale Feature Fusion With Adaptive Scale-Space Pyramid Network for Aerial Scene Recognition Using Remote Sensing Images

Remote Sensing is an area anthropogenic study undertaken worldwide. It has succeeded significantly in important applications such as climate monitoring, disaster prediction and land use planning. However, due to the diversity of scales, intra-class similarities, and complex scenes, the accurate recognition process remains challenging. Transformers' global attention mechanism helps them to overcome the limitations of CNNs' local receptive fields; however, they have drawback of increased computing complexity. To overcome such challenges, this work proposes an Adaptive Scale-Space Pyramid Network (ASSPN) for improved remote sensing image classification. The ASSPN architecture contains a learnable Gaussian pyramid module for multi-scale feature representation, a scale selection attention mechanism for dynamically weighing feature relevance, a cross-feature propagation module for fusion guided by uncertainty, and a complexity-aware adaptive pooling module for preserving semantic discriminative features. Experiments are performed three benchmark datasets such as EuroSAT, NWPU-RESISC-45, and MLRSNet. On these datasets, the ASSPN achieves state-of-the-art results with accuracies of 96.14%, 94.73%, and 95.42%, respectively. The obtained accuracy is outperforming previous CNN and transformer-based systems with significant margins. Furthermore, ASSPN is noise perturbation-resistant and shows generalization capability across a wide range of land-cover categories. Ablation studies established the complementary benefits of the core modules, while LIME-based explainability analysis confirmed the predicative trustworthiness of the model.

DFSNet-VLM: A Hybrid Frequency-Aware and Vision-Language Framework for Remote Sensing Scene Classification and Semantic Image Explanation

Remote sensing has always been an area of interest for researchers due to its significance in Earth monitoring, which supports proper future planning for agriculture, construction, reforestation, and climate change. Transformer architecture achieves significant performance in remote sensing image classification; however, they come with the trade-off of higher computational complexity. In this paper, we propose a novel deep learning framework, DFSNet-VLM—Cross Domain Fusion based Texture-Sensitive Dual Stream Network — for high-precision remote sensing scene understanding. The proposed framework includes a classification model, “DFSNet,” that improves feature representation by employing both spatial and frequency domain features, which ultimately help detect global patterns and textures alongside local features. The model also promotes information exchange between both streams to complement one type of features with respect to the other by integrating cross-domain fusion blocks at multiple stages. Additionally, a pretrained VLM model, “BLIP-2,” is integrated to provide semantic descriptions of classified images. Bayesian optimization is applied to fine tune hyperparameters, reducing overfitting and improving model performance. The proposed model is evaluated on six diverse publicly available datasets and achieves improved accuracies of 97.13% on MLRSNet, 94.67% on NWPU-RESISC-45, 98.00% on EuroSAT, 92.25% on GeoSceneNet16k, 98.25% on cloud, and 96.03% on the Bijie-landslide dataset, respectively. Detailed ablation studies, comparative analysis, and Grad-CAM++-based model explainability demonstrate that the proposed model is generalizable and scalable, and that it achieves improved accuracy. In addition, the proposed model can be easily implemented in a real-time environment for diverse applications. The trained model's links are available in the data availability section.

A Novel Reduced-Ripple Torque Control Technique for Electric Vehicle Switched Reluctance Motors Using an Adaptive Fuzzy Neural Network

This paper presents a novel torque control technique for SRMs using an adaptive fuzzy neural network (AFNN); the proposed technique allows for a precise current profiling with a superior ability of torque ripple reduction, even compared to torque sharing function (TSF) strategies. The proposed AFNN involves a four-layer neural network based on fuzzy logic; the torque reference and torque error are the inputs of FNN; the output is the reference commanded current; the parameters are optimized by a multi-objective Aquila Optimizer (AO) algorithm with only 16 parameters trained at a single operating point, generalizing across the full speed range. Moreover, the switching angles of the tested 12/8 SRM prototype are optimized; thereby contributing to improved torque quality and overall operational performance. Furthermore, the experimental and simulation results are achieved compared to TSF strategies; they reveal the superiority of proposed torque control technique over the entire speed range and different loading levels; it shows significant torque ripple reductions compared to TSF strategies, the proposed AFNN technique shows 52%, 83%, 89% reduction ratios of torque ripple for low-speed and light loads, medium-speed and heavy loads, and high-speed and light loads, respectively. Besides, it has fast dynamics, a simple structure, and a good generalization ability for several industrial applications, including EVs.

Spectroscopic Characterization of Plasmoid Properties During Pellet Fueling in W7-X

This study utilizes a spectroscopic approach to investigate the properties of plasmoids that are formed during the process of cryogenic hydrogen pellet fueling in the Wendelstein 7-X (W7-X) stellarator. An analysis of the Balmer series emissions was conducted using a diagnostic that was installed during the 2024 operational campaign. Electron temperature, density, and plasma beta (b) values can be inferred from the emissions of radiation from the ablation plasmoid. These values are essential for validating pellet ablation models and, in the future, optimizing fueling strategies in steady-state fusion devices.

Obicetrapib for dyslipidemia with or without cardiovascular risk: A GRADE ‐assessed meta‐analysis of randomized trials with trial sequential evidence

Abstract Aims Obicetrapib, an oral cholesteryl ester transfer protein ( CETP ) inhibitor, has demonstrated potent LDL ‐C lowering in recent phase 2/3 trials. We evaluated Obicetrapib (1, 2.5, 5, and 10 mg) efficacy and safety in adults with dyslipidemia, with or without atherosclerotic cardiovascular disease ( ASCVD ) risk. Materials and Methods We performed a meta‐analysis of randomized controlled trials ( RCTs ) identified through PubMed , Cochrane, Scopus, and Web of Science up to June 2025. Dichotomous outcomes were analyzed as risk ratios ( RRs ) and continuous outcomes as mean differences ( MDs ), both with 95% confidence intervals ( CIs ). PROSPERO ID : CRD420251107076.Results Six RCTs including 3399 patients were analysed. Compared with placebo, Obicetrapib significantly reduced LDL ‐C at 8–12 weeks ( MD –27.66 mg/ dL (−26.96%); 95% CI –33.62 to −21.70; p  < 0.0001) and non‐ HDL ‐C ( MD –35.41 mg/ dL (−28.08%); 95% CI –39.42 to −31.39; p  < 0.0001). It also increased HDL ‐C ( MD 70.85 mg/ dL (141.7%); 95% CI 62.56–79.15; p  < 0.0001) and improved achievement of LDL ‐C targets: <55 mg/ dL ( RR 6.42; 95% CI 5.15–8.01), <70 mg/ dL ( RR 2.56; p  < 0.0001), and < 100 mg/ dL ( RR 1.34; p  < 0.0001). No significant differences were found in total adverse events ( p  = 0.41) or serious adverse events ( p  = 0.37). Conclusion Obicetrapib provides substantial improvements in lipid parameters with a favourable short‐term adverse events rate. These results support its role as a potential adjunctive lipid‐lowering agent irrespective of ASCVD risk. Longer‐term trials are warranted to confirm its durability, cardiovascular outcomes, and safety.

Efficacy and Safety of OnabotulinumtoxinA for the Treatment of Platysma Prominence: A Systematic Review and Meta‐Analysis of Randomized Controlled Trials

ABSTRACT Background Platysma prominence (PP) is a common aesthetic concern associated with aging, leading to visible neck bands and loss of jawline definition. OnabotulinumtoxinA has emerged as a minimally invasive treatment; however, data from randomized controlled trials (RCTs) remain fragmented. Aims To systematically evaluate the efficacy and safety of onabotulinumtoxinA for treating moderate to severe PP through meta‐analysis of RCTs. Methods PubMed, Cochrane CENTRAL, and ClinicalTrials.gov were searched from inception to March 2025 for RCTs comparing onabotulinumtoxinA with placebo in adults with PP. Two reviewers independently extracted data and assessed bias using the Cochrane RoB 2.0 tool. The primary outcomes were ≥ 1‐grade and ≥ 2‐grade improvement on the Clinician (C‐APPS) and Participant (P‐APPS) Allergan Platysma Prominence Scales. Secondary outcomes included patient satisfaction and treatment‐emergent adverse events (TEAEs). Random‐effects meta‐analysis was used to estimate pooled risk ratios (RRs) with 95% confidence intervals (CIs). Results Three RCTs ( n  = 912) were included. OnabotulinumtoxinA significantly increased ≥ 1‐grade (RR = 4.11; 95% CI, 3.60–4.69) and ≥ 2‐grade (RR = 1.83; 95% CI, 1.54–2.17) improvements compared to placebo. Patient satisfaction was higher in the treatment group (RR = 5.55; 95% CI, 4.15–7.43). The incidence of TEAEs was similar between groups (RR = 0.95; 95% CI, 0.76–1.20), with most being mild and transient. Conclusions OnabotulinumtoxinA is an effective and well‐tolerated minimally invasive option for improving platysma prominence, offering significant aesthetic and patient‐reported benefits without increasing adverse effects.

Combining Robotics and Microbiome Science to Treat Unusual Post‐Herpetic Neuralgias Resembling Complex Regional Pain Syndrome