Search Results (279)

Hypersonic unmanned flight vehicles have complex dynamic characteristics, such as nonlinearity, strong coupling, multiple constraints, and uncertainty. Operating in highly complex flight environments, hypersonic unmanned flight vehicles must not only contend with uncertainties and disturbances such as parameter perturbations and noise but also deal with complex task scenarios such as interception and no-fly zone avoidance. These factors collectively pose great challenges on the control performance of the vehicle. To address the challenges of trajectory tracking for the vehicles under complex constraints, this paper proposes a trajectory tracking control method based on model predictive control (MPC). Firstly, a nonlinear dynamic model for hypersonic unmanned flight vehicles is established. Then, a robust model predictive controller is designed and the optimal control law is derived to address the trajectory tracking control problem under complex constraints such as parameter perturbations. Finally, simulation experiments are designed under the conditions of aerodynamic parameter changes in the longitudinal plane and lateral no-fly zone avoidance. The simulation results demonstrate that the vehicle is capable of accurately and rapidly tracking the reference despite aerodynamic parameter perturbations and large-scale lateral maneuvers, thereby validating the effectiveness of the controller. Full article

With the advancements in deep learning methods, AI systems now perform at the same or higher level than human intelligence in many complex real-world problems. The data and algorithmic opacity of deep learning models, however, make the task of comprehending the input data information, the model, and model’s decisions quite challenging. This lack of transparency constitutes both a practical and an ethical issue. For the present study, it is a major drawback to the deployment of deep learning methods mandated with detecting patterns and prognosticating Alzheimer’s disease. Many approaches presented in the AI and medical literature for overcoming this critical weakness are sometimes at the cost of sacrificing accuracy for interpretability. This study is an attempt at addressing this challenge and fostering transparency and reliability in AI-driven healthcare solutions. The study explores a few commonly used perturbation-based interpretability (LIME) and gradient-based interpretability (Saliency and Grad-CAM) approaches for visualizing and explaining the dataset, models, and decisions of MRI image-based Alzheimer’s disease identification using the diagnostic and predictive strengths of an ensemble framework comprising Convolutional Neural Networks (CNNs) architectures (Custom multi-classifier CNN, VGG-19, ResNet, MobileNet, EfficientNet, DenseNet), and a Vision Transformer (ViT). The experimental results show the stacking ensemble achieving a remarkable accuracy of 98.0% while the hard voting ensemble reached 97.0%. The findings present a valuable contribution to the growing field of explainable artificial intelligence (XAI) in medical imaging, helping end users and researchers to gain deep understanding of the backstory behind medical image dataset and deep learning model’s decisions. Full article

To address the challenges of low optimization efficiency and premature convergence in existing algorithms for unmanned aerial vehicle (UAV) 3D path planning under complex operational constraints, this study proposes an enhanced honey badger algorithm (LRMHBA). First, a three-dimensional terrain model incorporating threat sources and UAV constraints is constructed to reflect the actual operational environment. Second, LRMHBA improves global search efficiency by optimizing the initial population distribution through the integration of Latin hypercube sampling and an elite population strategy. Subsequently, a stochastic perturbation mechanism is introduced to facilitate the escape from local optima. Furthermore, to adapt to the evolving exploration requirements during the optimization process, LRMHBA employs a differential mutation strategy tailored to populations with different fitness values, utilizing elite individuals from the initialization stage to guide the mutation process. This design forms a two-population cooperative mechanism that enhances the balance between exploration and exploitation, thereby improving convergence accuracy. Experimental evaluations on the CEC2017 benchmark suite demonstrate the superiority of LRMHBA over 11 comparison algorithms. In the UAV 3D path planning task, LRMHBA consistently generated the shortest average path across three obstacle simulation scenarios of varying complexity, achieving the highest rank in the Friedman test. Full article

Objective: Mental fatigue (MF) induced by prolonged cognitive tasks poses significant risks to postural stability, yet its effects on multi-sensory integration remain poorly understood. Method: This study investigated how MF alters sensory reweighting and postural control in 27 healthy young males. A 45 min incongruent Stroop task was employed to induce MF, validated via subjective Visual Analog Scale (VAS) scores and psychomotor vigilance tests. Postural stability was assessed under four sensory perturbation conditions (O-H: no interference; C-H: visual occlusion; O-S: proprioceptive perturbation; C-S: combined perturbations) using a Kistler force platform. Center of pressure (COP) signals were analyzed through time-domain metrics, sample entropy (SampEn), and Discrete Wavelet Transform (DWT) to quantify energy distributions across sensory-related frequency bands (visual: 0–0.1 Hz; vestibular: 0.1–0.39 Hz; cerebellar: 0.39–1.56 Hz; proprioceptive: 1.56–6.25 Hz). Results: MF significantly reduced proprioceptive energy contributions (p < 0.05) while increasing vestibular reliance under O-S conditions (p < 0.05). Time-domain metrics showed no significant changes in COP velocity or displacement, but SampEn decreased under closed-eye conditions (p < 0.001), indicating reduced postural adaptability. DWT analysis highlighted MF’s interaction with visual occlusion, altering cerebellar and proprioceptive energy dynamics (p < 0.01). Conclusion: These findings demonstrate that MF disrupts proprioceptive integration, prompting compensatory shifts toward vestibular and cerebellar inputs. The integration of nonlinear entropy and frequency-domain analyses advances methodological frameworks for fatigue research, offering insights into real-time sensor-based fatigue monitoring and balance rehabilitation strategies. This study underscores the hierarchical interplay of sensory systems under cognitive load and provides empirical evidence for optimizing interventions in high-risk occupational and clinical settings. Full article

Deep learning models are often perceived as black boxes, making it challenging to analyze the causal relationships between inputs and outputs. For this reason, the explainability of model learning has garnered increasing attention in recent years. Some previous studies proposed influence functions, which evaluate how the weighting of data impacts the model by mathematical analysis, thereby explaining how it realizes the data. This inspires us to suggest that when data in an upstream task is affected by varying levels of noise interference, it is practical to set up a downstream model to apply Taylor expansion in conjunction with the Hessian matrix to estimate perturbations that each data point cause in the model. Additionally, utilizing Integrated Gradients to compute the loss difference between the original data instances and a baseline instance which does not affect the model is powerful to yield a memorization matrix that allows researchers to observe the changes in model reasoning before and after noise interference, helping to analyze the causes of erroneous inference. Full article

Since the research of $\mu$-synchronization helps to explore how complex networks (CNs) work together to produce complex behaviors, the $\mu$-synchronization task for uncertain time-delayed CNs is studied in our work. Especially, bounded external perturbations and variable delays of unknown bounds containing coupling delays, internal delays, and pulse delays are all taken into consideration, making the model more general. Through the $\mu$-stable theory together with the hybrid impulsive control technique, the problems caused by uncertain inner couplings, time-varying delays, and perturbations can be solved, and novel synchronization criteria are gained for the $\mu$-synchronization of the considered CNs. Different from traditional models, it is not necessary for the coupling matrices to meet the zero-row-sum condition, and the control protocol relaxes the constraint of time delays on impulse intervals. Moreover, numerical experiments and image encryption algorithms are carried out to verify our theoretical results’ effectiveness. Full article

Machine learning (ML) promises advances in automation and threat detection for the future generations of communication networks. However, new threats are introduced, as adversaries target ML systems with malicious data. Adversarial attacks on tree-based ML models involve crafting input perturbations that exploit non-smooth decision boundaries, causing misclassifications. These so-called evasion attacks are imperceptible, as they do not significantly alter the input data distribution and have been shown to degrade the performance of tree-based models across various tasks. Adversarial training and genetic algorithms have been proposed as potential defenses against these attacks. In this paper, we explore the robustness of tree-based models for network intrusion detection systems. This study evaluates an optimization approach inspired by genetic algorithms to generate adversarial samples and studies the impact of adversarial training on the accuracy of attack detection. This paper exposed random forest and extreme gradient boosting classifiers to various adversarial samples generated from communication network-related CIC-IDS2019 and 5G-NIDD datasets. The results indicate that the improvements of robustness to adversarial attacks come with a cost to the accuracy of the network intrusion detection models. These costs can be optimized with intelligent, use case-specific feature engineering. Full article

Audio tagging, as a fundamental task in acoustic signal processing, has demonstrated significant advances and broad applications in recent years. Spiking Neural Networks (SNNs), inspired by biological neural systems, exploit event-driven computing paradigms and temporal information processing, enabling superior energy efficiency. Despite the increasing adoption of SNNs, the potential of event-driven encoding mechanisms for audio tagging remains largely unexplored. This work presents a pioneering investigation into event-driven encoding strategies for SNN-based audio tagging. We propose the SATRN (Spiking Audio Tagging Robust Network), a novel architecture that integrates temporal–spatial attention mechanisms with membrane potential residual connections. The network employs a dual-stream structure combining global feature fusion and local feature extraction through inverted bottleneck blocks, specifically designed for efficient audio processing. Furthermore, we introduce an event-based encoding approach that enhances the resilience of Spiking Neural Networks to disturbances while maintaining performance. Our experimental results on the Urbansound8k and FSD50K datasets demonstrate that the SATRN achieves comparable performance to traditional Convolutional Neural Networks (CNNs) while requiring significantly less computation time and showing superior robustness against noise perturbations, making it particularly suitable for edge computing scenarios and real-time audio processing applications. Full article

Due to the challenge of acquiring abundant labeled samples, semi-supervised change detection (SSCD) approaches are becoming increasingly popular in tackling CD tasks with limited labeled data. Despite their success, these methods tend to come with complex network architectures or cumbersome training procedures, which also ignore the domain gap between the labeled data and unlabeled data. Differently, we hypothesize that diverse perturbations are more favorable to exploit the potential of unlabeled data. In light of this spirit, we propose a novel SSCD approach based on Weak–strong Augmentation and Class-balanced Sampling (WACS-SemiCD). Specifically, we adopt a simple mean-teacher architecture to deal with labeled branch and unlabeled branch separately, where supervised learning is conducted on the labeled branch, while weak–strong consistency learning (e.g., sample perturbations’ consistency and feature perturbations’ consistency) is imposed for the unlabeled. To improve domain generalization capacity, an adaptive CutMix augmentation is proposed to inject the knowledge from the labeled data into the unlabeled data. A class-balanced sampling strategy is further introduced to mitigate class imbalance issues in CD. Particularly, our proposed WACS-SemiCD achieves competitive SSCD performance on three publicly available CD datasets under different labeled settings. Comprehensive experimental results and systematic analysis underscore the advantages and effectiveness of our proposed WACS-SemiCD. Full article

Quantifying the water budget in the Great Salt Lake (GSL) basin is a nontrivial task, especially under a changing climate that contributes to increasing temperatures and a shift towards more rainfall and less snowfall. This study examines the potential impacts of temperature thresholds on the water budget in the GSL, emphasizing the influence on snowmelt, evapotranspiration (ET), and runoff under varying climate warming scenarios. Current hydrological models such as the Variable Infiltration Capacity (VIC) model use a universal temperature threshold to partition snowfall and rainfall across different regions. Previous studies have argued that there is a wide range of thresholds for partitioning rainfall and snowfall across the globe. However, there is a clear knowledge gap in quantifying water budget components in the Great Salt Lake (GSL) basin corresponding to varying temperature thresholds for separating rainfall and snowfall under the present and future climates. To address this gap, the study applied temperature thresholds derived from observation-based data available from National Center for Environmental Prediction (NCEP) to the VIC model. We also performed a suite of hydrological experiments to quantify the water budget of the Great Salt Lake basin by perturbing temperature thresholds and climate forcing. The results indicate that higher temperature thresholds contribute to earlier snowmelt, reduced snowpack, and lower peak runoff values in the early spring that are likely due to increased ET before peak runoff periods. The results show that the GSL undergoes higher snow water equivalent (SWE) values during cold seasons due to snow accumulation and lower values during warm seasons as increased temperatures intensify ET. Projected climate warming may result in further reductions in SWE (~71%), increased atmospheric water demand, and significant impacts on water availability (i.e., runoff reduced by ~20%) in the GSL basin. These findings underscore the potential challenges that rising temperatures pose to regional water availability. Full article

Path planning is of great research significance as it is key to affecting the efficiency and safety of mobile robot autonomous navigation task execution. The traditional gray wolf optimization algorithm is widely used in the field of path planning due to its simple structure, few parameters, and easy implementation, but the algorithm still suffers from the disadvantages of slow convergence, ease of falling into the local optimum, and difficulty in effectively balancing exploration and exploitation in practical applications. For this reason, this paper proposes a multi-strategy improved gray wolf optimization algorithm (MSIAR-GWO) based on reinforcement learning. First, a nonlinear convergence factor is introduced, and intelligent parameter configuration is performed based on reinforcement learning to solve the problem of high randomness and over-reliance on empirical values in the parameter selection process to more effectively coordinate the balance between local and global search capabilities. Secondly, an adaptive position-update strategy based on detour foraging and dynamic weights is introduced to adjust the weights according to changes in the adaptability of the leadership roles, increasing the guiding role of the dominant individual and accelerating the overall convergence speed of the algorithm. Furthermore, an artificial rabbit optimization algorithm bypass foraging strategy, by adding Brownian motion and Levy flight perturbation, improves the convergence accuracy and global optimization-seeking ability of the algorithm when dealing with complex problems. Finally, the elimination and relocation strategy based on stochastic center-of-gravity dynamic reverse learning is introduced for the inferior individuals in the population, which effectively maintains the diversity of the population and improves the convergence speed of the algorithm while avoiding falling into the local optimal solution effectively. In order to verify the effectiveness of the MSIAR-GWO algorithm, it is compared with a variety of commonly used swarm intelligence optimization algorithms in benchmark test functions and raster maps of different complexities in comparison experiments, and the results show that the MSIAR-GWO shows excellent stability, higher solution accuracy, and faster convergence speed in the majority of the benchmark-test-function solving. In the path planning experiments, the MSIAR-GWO algorithm is able to plan shorter and smoother paths, which further proves that the algorithm has excellent optimization-seeking ability and robustness. Full article

Purpose: Sex classification is a major benchmark of previous work in learning on the structural connectome, a naturally occurring brain graph that has proven useful for studying cognitive function and impairment. While graph neural networks (GNNs), specifically graph convolutional networks (GCNs), have gained popularity lately for their effectiveness in learning on graph data, achieving strong performance in adult sex classification tasks, their application to pediatric populations remains unexplored. We seek to characterize the capacity for GNN models to learn connectomic patterns on pediatric data through an exploration of training techniques and architectural design choices. Methods: Two datasets comprising an adult BRIGHT dataset (N = 147 Hodgkin’s lymphoma survivors and N = 162 age similar controls) and a pediatric Human Connectome Project in Development (HCP-D) dataset (N = 135 healthy subjects) were utilized. Two GNN models (GCN simple and GCN residual), a deep neural network (multi-layer perceptron), and two standard machine learning models (random forest and support vector machine) were trained. Architecture exploration experiments were conducted to evaluate the impact of network depth, pooling techniques, and skip connections on the ability of GNN models to capture connectomic patterns. Models were assessed across a range of metrics including accuracy, AUC score, and adversarial robustness. Results: GNNs outperformed other models across both populations. Notably, adult GNN models achieved 85.1% accuracy in sex classification on unseen adult participants, consistent with prior studies. The extension of the adult models to the pediatric dataset and training on the smaller pediatric dataset were sub-optimal in their performance. Using adult data to augment pediatric models, the best GNN achieved comparable accuracy across unseen pediatric (83.0%) and adult (81.3%) participants. Adversarial sensitivity experiments showed that the simple GCN remained the most robust to perturbations, followed by the multi-layer perceptron and the residual GCN. Conclusions: These findings underscore the potential of GNNs in advancing our understanding of sex-specific neurological development and disorders and highlight the importance of data augmentation in overcoming challenges associated with small pediatric datasets. Further, they highlight relevant tradeoffs in the design landscape of connectomic GNNs. For example, while the simpler GNN model tested exhibits marginally worse accuracy and AUC scores in comparison to the more complex residual GNN, it demonstrates a higher degree of adversarial robustness. Full article

Adversarial attacks deliberately modify deep learning inputs, mislead models, and cause incorrect results. Previous adversarial attacks on sentiment analysis models have demonstrated success in misleading these models. However, most existing attacks in sentiment analysis have applied a generalized approach to input modifications, without considering the characteristics and objectives of the different analysis levels. Specifically, for aspect-based sentiment analysis, there is a lack of attack methods that modify inputs in accordance with the evaluated aspects. Consequently, unnecessary modifications are made, compromising the input semantics, making the changes more detectable, and avoiding the identification of new vulnerabilities. In previous work, we proposed a model to generate adversarial examples in particular for aspect-based sentiment analysis. In this paper, we assess the effectiveness of our adversarial example model in negatively impacting aspect-based model results while maintaining high levels of semantic inputs. To conduct this evaluation, we propose diverse adversarial attacks across different dataset domains, target architectures, and consider distinct levels of victim model knowledge, thus obtaining a comprehensive evaluation. The obtained results demonstrate that our approach outperforms existing attack methods in terms of accuracy reduction and semantic similarity, achieving a 65.30% reduction in model accuracy with a low perturbation ratio of 7.79%. These findings highlight the importance of considering task-specific characteristics when designing adversarial examples, as even simple modifications to elements that support task classification can successfully mislead models. Full article

Treadmill running simulates various conditions, including flat, uphill, and downhill gradients, making it useful for training and rehabilitation. This study aimed to examine how incline and decline treadmill running affect local dynamic stability of individual running movement components that cooperatively contribute to achieving the running tasks. Principal component analysis (PCA) was used to decompose movement components, termed principal movements (PMs), from kinematic marker data collected from 19 healthy recreational runners (9 females and 10 males, 23.6 ± 3.7 years) during treadmill running at 10 km/h across different gradients (−6, −3, 0, +3, +6 degrees). The largest Lyapunov exponent (LyE) of individual PM positions (higher LyE = greater instability) was analyzed using repeated-measures ANOVA to assess treadmill gradient effects across PMs. The results showed that the effects of treadmill gradient appear in PM₃, which corresponds to the mid-stance phase of the gait cycle. Specifically, decline treadmill running significantly decreased local dynamic stability (greater LyE) compared to equivalent incline conditions (p ≤ 0.005). These findings suggest that decline treadmill running should be used cautiously in rehabilitation settings due to its potential to reduce an ability to control and respond to small perturbations, thereby increasing the risk of instability during the weight-bearing support phase of gait. Full article

Determining the values of various properties for new bio-inks for 3D printing is a very important task in the design of new materials. For this purpose, a large number of experimental works have been consulted, and a database with more than 1200 bioprinting tests has been created. These tests cover different combinations of conditions in terms of print pressure, temperature, and needle values, for example. These data are difficult to deal with in terms of determining combinations of conditions to optimize the tests and analyze new options. The best model demonstrated a specificity (Sp) of 88.4% and a sensitivity (Sn) of 86.2% in the training series while achieving an Sp of 85.9% and an Sn of 80.3% in the external validation series. This model utilizes operators based on perturbation theory to analyze the complexity of the data. For comparative purposes, neural networks have been used, and very similar results have been obtained. The developed tool could easily be applied to predict the properties of bioprinting assays in silico. These findings could significantly improve the efficiency and accuracy of predictive models in bioprinting without resorting to trial-and-error tests, thereby saving time and funds. Ultimately, this tool may help pave the way for advances in personalized medicine and tissue engineering. Full article