Adaptive Weight Research Articles

Adaptive weight adjustment technology for reinforcement learning reward indicators based on the new-type power system

Current RL-based scheduling methods struggle with designing effective reward functions that balance cost reduction, safety and renewable energy integration. To tackle these issues, this paper proposes a novel hybrid optimization framework that combines Particle Swarm Optimization and Reinforcement Learning (PSO-RL), focusing on three key aspects: (1) Minimizing Generation Costs: Systematically reducing operational costs for economic efficiency. (2) Enhancing System Safety: Improving reliability by incorporating operational constraints. (3) Increasing Renewable Energy Proportion: Emphasizing the integration of renewable energy sources. The comprehensive mathematical model encapsulates intricate constraints and multi-objective characteristics of the new-type power system, featuring reward functions tailored for dynamic responsiveness. Empirical validations using the Grid2Op scheduler affirm the efficacy of our framework. Results highlight superior convergence speed and optimization precision of the PSO-RL approach over traditional RL methods. In summary, this study advances scheduling methodologies for the new-type power system, fostering renewable integration and smart grid technology evolution.

Relevance of body weight adaptation and modern obesity-defining parameters in the analysis of isokinetic trunk strength in people with obesity - A retrospective analysis.

Pathologic values of body mass index (BMI), body weight, and waist circumference correlate with higher absolute and lower relative trunk strength. Whether waist-to-height ratio (WHtR) is appropriate for showing trunk strength differences in people with obesity and whether a continuous linear relationship exists between the increase in obesity and trunk strength is unknown. This retrospective cross-sectional study included 1174 subjects (1114 men and 60 women). Measured values included body weight, height, waist circumference, WHtR, BMI, and both absolute and body weight-adapted trunk flexor/extensor strength. Statistical analyses included t-tests, Welch tests, Pearson correlations, mixed-linear, and nonlinear regression analyses. Positive correlations with absolute trunk strength were found in subjects without obesity for all anthropometric parameters except WHtR. Weaker positive and partly negative correlation and linear regression coefficients were found in subjects with obesity. Nonlinear relationships were found between age, BMI, WHtR, and absolute respective body weight-adapted trunk strength. The relationship between obesity-defining measures/ indices and trunk strength is non-linear. Increasing BMI, waist circumference, or WHtR above cut-off values known from cardiovascular research is linked to a decrease or weaker increase in trunk strength. Body weight adaptation is recommended to avoid misinterpretation of apparently good absolute trunk strength values in people with obesity.

Knowledge distillation with adapted weight

Although large models have shown a strong capacity to solve large-scale problems in many areas including natural language and computer vision, their voluminous parameters are hard to deploy in a real-time system due to computational and energy constraints. Addressing this, knowledge distillation through Teacher-Student architecture offers a sustainable pathway to compress the knowledge of large models into more manageable sizes without significantly compromising performance. To enhance the robustness and interpretability of this framework, it is critical to understand how individual training data impact model performance, which is an area that remains underexplored. We propose the Knowledge Distillation with Adaptive Influence Weight (KD-AIF) framework which leverages influence functions from robust statistics to assign weights to training data, grounded in the four key SAFE principles: Sustainability, Accuracy, Fairness, and Explainability. This novel approach not only optimizes distillation but also increases transparency by revealing the significance of different data. The exploration of various update mechanisms within the KD-AIF framework further elucidates its potential to significantly improve learning efficiency and generalization in student models, marking a step toward more explainable and deployable Large Models. KD-AIF is effective in knowledge distillation while also showing exceptional performance in semi-supervised learning with outperforms existing baselines and methods in multiple benchmarks (CIFAR-100, CIFAR-10-4k, SVHN-1k, and GLUE).

Federated Adaptive Causal Estimation (FACE) of Target Treatment Effects

Federated learning of causal estimands may greatly improve estimation efficiency by leveraging data from multiple study sites, but robustness to heterogeneity and model misspecifications is vital for ensuring validity. We develop a Federated Adaptive Causal Estimation (FACE) framework to incorporate heterogeneous data from multiple sites to provide treatment effect estimation and inference for a flexibly specified target population of interest. FACE accounts for site-level heterogeneity in the distribution of covariates through density ratio weighting. To safely incorporate source sites and avoid negative transfer, we introduce an adaptive weighting procedure via a penalized regression, which achieves both consistency and optimal efficiency. Our strategy is communication-efficient and privacy-preserving, allowing participating sites to share summary statistics only once with other sites. We conduct both theoretical and numerical evaluations of FACE and apply it to conduct a comparative effectiveness study of BNT162b2 (Pfizer) and mRNA-1273 (Moderna) vaccines on COVID-19 outcomes in U.S. veterans using electronic health records from five VA regional sites. We show that compared to traditional methods, FACE meaningfully increases the precision of treatment effect estimates, with reductions in standard errors ranging from 26 % to 67 % .

Spatio-Temporal Joint Trajectory Planning for Autonomous Vehicles Based on Improved Constrained Iterative LQR.

With advancements in autonomous driving technology, the coupling of spatial paths and temporal speeds in complex scenarios becomes increasingly significant. Traditional sequential decoupling methods for trajectory planning are no longer sufficient, emphasizing the need for spatio-temporal joint trajectory planning. The Constrained Iterative LQR (CILQR), based on the Iterative LQR (ILQR) method, shows obvious potential but faces challenges in computational efficiency and scenario adaptability. This paper introduces three key improvements: a segmented barrier function truncation strategy with dynamic relaxation factors to enhance stability, an adaptive weight parameter adjustment method for acceleration and curvature planning, and the integration of the hybrid A* algorithm to optimize the initial reference trajectory and improve iterative efficiency. The improved CILQR method is validated through simulations and real-vehicle tests, demonstrating substantial improvements in human-like driving performance, traffic efficiency improvement, and real-time performance while maintaining comfortable driving. The experiment's results demonstrate a significant increase in human-like driving indicators by 16.35% and a 12.65% average increase in traffic efficiency, reducing computation time by 39.29%.

A Combination of Classification Robust Adaptive Kalman Filter with PPP-RTK to Improve Fault Detection for Integrity Monitoring of Autonomous Vehicles

Real-time integrity monitoring (IM) is essential for autonomous vehicle positioning, requiring high availability and manageable computational load. This research proposes using precise point positioning real-time kinematic (PPP-RTK) as the positioning method, combined with an improved classification adaptive Kalman filter (CAKF) for processing. PPP-RTK enhances IM availability by allowing undifferenced and uncombined observations, enabling individual observation exclusion during fault detection and exclusion (FDE). The CAKF reduces FDE computational load by using a robustness test instead of traditional FDE methods, improving precision and availability in protection level estimation. Epoch-wise weighting adjustments in the robustness test create a more accurate stochastic model, aided by an adaptive unit weight variance (UWV) calculated with a sliding window, achieving a 7–28% UWV reduction. Three test scenarios with up to four simultaneous faults in code and phase observations, ranging from 1 to 200 m and 0.4 to 20 m, respectively, demonstrated successful identification and de-weighting of faults, resulting in maximum positioning errors of 6 mm (horizontal) and 11 mm (vertical). The method reduced FDE computational load by 50–99.999% compared to other approaches.

SPATIAL MODELING OF MATERNAL HEALTH: GEOGRAPHICALLY WEIGHTED POISSON REGRESSION ON MATERNAL MORTALITY FACTORS

Data from the 2021 West Java Provincial Health Profile Report, accessed from the official website of the West Java Provincial Health Office, reveals a significant surge in maternal mortality cases, rising from 165 in 2020 to 460 in 2021. In support of efforts to reduce maternal mortality rates, this study investigates the contributing factors to this phenomenon across various districts in West Java Province. The data used is from the year 2021. This study aims to evaluate the effectiveness of Poisson regression, negative binomial regression, and Geographically Weighted Poisson Regression (GWPR) models in capturing the variability of maternal deaths in the study area for that year. A comprehensive analysis revealed that the distribution of maternal mortality fits the Poisson model, displaying significant spatial heterogeneity. Acknowledging this variability, the GWPR approach using an Adaptive Kernel Bisquare weighting was selected due to its capability to produce localized parameter estimates, which more accurately reflect the specific conditions of each location. The analyzed independent variables include the number of community health centers, coverage of antenatal services at the first (K1) and fourth (K4) visits, management of obstetric complications, and coverage of iron supplementation for pregnant women. Of the five variables, only three showed statistically significant effects; therefore, the study proceeded using these three variables. The results indicate that GWPR provides the best explanation for the variability in maternal mortality rates, with an adjusted R² value of 63.17% and a MAPE of 37.70%.

Speech Emotion Recognition Model Based on Joint Modeling of Discrete and Dimensional Emotion Representation

This paper introduces a novel joint model architecture for Speech Emotion Recognition (SER) that integrates both discrete and dimensional emotional representations, allowing for the simultaneous training of classification and regression tasks to improve the comprehensiveness and interpretability of emotion recognition. By employing a joint loss function that combines categorical and regression losses, the model ensures balanced optimization across tasks, with experiments exploring various weighting schemes using a tunable parameter to adjust task importance. Two adaptive weight balancing schemes, Dynamic Weighting and Joint Weighting, further enhance performance by dynamically adjusting task weights based on optimization progress and ensuring balanced emotion representation during backpropagation. The architecture employs parallel feature extraction through independent encoders, designed to capture unique features from multiple modalities, including Mel-frequency Cepstral Coefficients (MFCC), Short-term Features (STF), Mel-spectrograms, and raw audio signals. Additionally, pre-trained models such as Wav2Vec 2.0 and HuBERT are integrated to leverage their robust latent features. The inclusion of self-attention and co-attention mechanisms allows the model to capture relationships between input modalities and interdependencies among features, further improving its interpretability and integration capabilities. Experiments conducted on the IEMOCAP dataset using a leave-one-subject-out approach demonstrate the model’s effectiveness, with results showing a 1–2% accuracy improvement over classification-only models. The optimal configuration, incorporating the joint architecture, dynamic weighting, and parallel processing of multimodal features, achieves a weighted accuracy of 72.66%, an unweighted accuracy of 73.22%, and a mean Concordance Correlation Coefficient (CCC) of 0.3717. These results validate the effectiveness of the proposed joint model architecture and adaptive balancing weight schemes in improving SER performance.

A Line of Sight/Non Line of Sight Recognition Method Based on the Dynamic Multi-Level Optimization of Comprehensive Features.

With the advent of the 5G era, high-precision localization based on mobile communication networks has become a research hotspot, playing an important role in indoor emergency rescue in shopping malls, smart factory management and tracking, as well as precision marketing. However, in complex environments, non-line-of-sight (NLOS) propagation reduces the measurement accuracy of 5G signals, causing large deviations in position solving. In order to obtain high-precision position information, it is necessary to recognize the propagation state of the signal before distance measurement or angle measurement. In this paper, we propose a dynamic multi-level optimization of comprehensive features (DMOCF) network model for line-of-sight (LOS)/NLOS identification. The DMOCF model improves the expression ability of the deep model by adding a res2 module to the time delay neural network (TDNN), so that fine-grained feature information such as weak reflections or noise in the signal can be deeply understood by the model, enabling the network to realize layer-level feature processing by adding Squeeze and Excitation (SE) blocks with adaptive weight adjustment for each layer. A mamba module with position coding is added to each layer to capture the local patterns of wireless signals under complex propagation phenomena by extracting local features, enabling the model to understand the evolution of signals over time in a deeper way. In addition, this paper proposes an improved sand cat search algorithm for network parameter search, which improves search efficiency and search accuracy. Overall, this new network architecture combines the capabilities of local feature extraction, global feature preservation, and time series modeling, resulting in superior performance in the 5G channel impulse response (CIR) signal classification task, improving the accuracy of the model and accurately identifying the key characteristics of multipath signal propagation. Experimental results show that the NLOS/LOS recognition method proposed in this paper has higher accuracy than other deep learning methods.

Learning and Phase Tracking by Frequency and Weight Adaptation for Coupled Networks of Kuramoto Oscillators

Learning and Phase Tracking by Frequency and Weight Adaptation for Coupled Networks of Kuramoto Oscillators

A deep learning method based on multi-scale fusion for noise-resistant coal-gangue recognition

Coal-gangue recognition technology plays an important role in the intelligent realization of integrated working faces and coal quality improvement. However, the existing methods are easily affected by high dust, noise, and other disturbances, resulting in unstable recognition results that make it difficult to meet the needs of industrial applications. To realize accurate recognition of coal-gangue in noisy environments, this paper proposes an end-to-end multi-scale feature fusion convolutional neural network (MCNN-BILSTM) based gangue recognition method, which can automatically learn and fuse complementary information from multiple signal components of vibration signals. It combines traditional filtering methods and the idea of multi-scale learning, which can expand the breadth and depth of the feature learning process. the breadth and depth of the feature learning process. Moreover, to strengthen the expression of key features, a feature weighting method based on the attention mechanism is combined to give adaptive weights to different features. Finally, the experimental platform of a tail beam of coal-gangue impact hydraulic support is built, and several comparative experiments are carried out. The comprehensive comparison experiments show that the method shows strong adaptability, robustness, and noise resistance under various complex noise environments, and is suitable for complex practical industrial sites.

Enhanced clinical photoacoustic vascular imaging through a skin localization network and adaptive weighting

Enhanced clinical photoacoustic vascular imaging through a skin localization network and adaptive weighting

Enhancing sidewalk accessibility assessment for wheelchair users: An adaptive weighting fuzzy-based approach.

To reach a destination within the community, it is crucial that wheelchair users possess the ability to plan, execute, and acquire knowledge of routes in a safe and efficient manner. While numerous methods have been introduced for assessing the accessibility of sidewalks, existing studies often overlook the variations in the perception of the accessibility of long segments based on each wheelchair user's capabilities. Extended distances may lead to increased fatigue, impacting the ability of individuals with mobility disabilities to navigate sidewalks comfortably and independently. In this paper, we propose an adaptive weighting method, effectively addressing the accessibility assessment of sidewalks by considering more specifically the impact of sidewalk length. The results underscore the significant impact of sidewalk length on mobility, delineating varying accessibility indices in long sidewalk segments, and offering a more realistic evaluation of accessibility based on wheelchair users' perceptions. For validation purposes, the proposed model was implemented in a personalized routing tool called MobiliSIG and compared with the conventional fuzzy model provided by the tool for accessibility assessment through a case study in Quebec City. The results demonstrated improved routing outcomes compared to previous methods, showcasing the effectiveness of our model in enhancing sidewalk accessibility assessment.

Detect influential points of feature rankings.

Feature rankings are crucial in bioinformatics but can be distorted by influential points (IPs), which are often overlooked. This study aims to investigate the impact of IPs on feature rankings and propose IPs detection method METHOD: We use a leave-one-out approach to assess each case's influence on feature rankings by comparing rank changes after its removal. The rank changes are measured by a novel rank comparison method that involves using adaptive top-prioritized weights that are adjustable to the distribution of rank changes. Our IP detection method was evaluated on several public datasets. Our method identified potential IPs in several TCGA gene expression datasets, revealing that IPs can severely distort feature rankings. These rank changes can ultimately affect subsequent analyses such as enriched pathways, suggesting the necessity of IPs detection when deriving feature rankings. IPs significantly impact feature rankings and subsequent analyses; routine IP detection is necessary yet underutilized. Our method is available in the R package findIPs.

Intelligent Data-Driven Task Offloading Framework for Internet of Vehicles Using Edge Computing and Reinforcement Learning

Introduction: The Internet of Vehicles (IoV) was enabled through innovative developments featuring advanced automotive networking and communication to fulfill the need for real-time applications that are latency-sensitive, such as autonomous driving and emergency management. Given that the servers were much farther away from the actual site of operation, traditional cloud computing faced huge delays in processing. Mobile Edge Computing (MEC) resolved this challenge by enabling localized data processing, reducing latency and enhancing resource utilization.Methods: This study proposed an Efficient Mobile Edge Computing-based Internet of Vehicles Task Offloading Framework (EMEC-IoVTOF). The framework integrated deep reinforcement learning (DRL) to optimize task offloading decisions, focusing on minimizing latency and energy consumption while accounting for bandwidth and computational constraints. Offloading costs were calculated using mathematical modeling and further optimized through Particle Swarm Optimization (PSO). An adaptive inertia weight mechanism was implemented to avoid local optimization and enhance task allocation decisions.Result: The proposed framework was thus proved effective for any latency reduction and energy consumption optimization in efficiently improving the overall system performance. DRL and MEC together facilitate scalability in task distribution by ensuring robust performance in dynamic vehicular environments. Integration with PSO further enhances the decision-making process and makes the system highly adaptable to dynamic task demands and network conditions.Discussion:The findings highlighted the potential of EMEC-IoVTOF to address key challenges in IoV systems, including latency, energy efficiency, and bandwidth utilization. Future research could explore real-world deployment and adaptability to complex vehicular scenarios, further validating its scalability and reliability.

A robust bearing fault diagnosis method based on ensemble learning with adaptive weight selection

A robust bearing fault diagnosis method based on ensemble learning with adaptive weight selection

Feature selection through adaptive sparse learning for scene recognition

Scene recognition is an important and challenging task in the field of computer vision. Current research typically focuses on local features in scene images by utilizing pretrained convolutional neural networks (CNN) to obtain a feature dictionary. However, these local features often contain similar features to other scene categories, leading to feature confusion and subsequently low accuracy in scene recognition. Therefore, focusing on local features for scene recognition is controversial. In contrast to existing works, we consider extracting common features within the same category from scene images and using them as discriminative features between different categories. We propose a scene recognition method based on adaptive sparse learning feature selection. This method leverages sparse learning to distinguish the contribution of each feature in the deep features to the classification task, aiming to construct a salient feature dictionary that describes the importance of features in scene images. Subsequently, an adaptive weight optimization method is employed through alternating updates to automatically adjust feature weights, enabling the selection of common features within the same scene category from the compact dictionary. Experimental results on multiple benchmark scene recognition datasets demonstrate that our proposed method is superior to state-of-the-art methods.

Adaptive Inertia Weight with Transient Search Optimization Based Feature Selection for Intrusion Detection in Internet of Things

Adaptive Inertia Weight with Transient Search Optimization Based Feature Selection for Intrusion Detection in Internet of Things

A dual-adaptive stochastic reinforcement chimp optimization algorithm for fire detection and multidimensional problem solving

Chimp optimization algorithm (CHOA) is a recently developed nature-inspired technique that mimics the swarm intelligence of chimpanzee colonies. However, the original CHOA suffers from slow convergence and a tendency to reach local optima when dealing with multidimensional problems. To address these limitations, we propose TASR-CHOA, a twofold adaptive stochastic reinforced variant. The TASR-CHOA algorithm integrates two novel methodologies: a stochastic approach to improve the speed at which convergence is achieved and a dual adaptive weighting approach to optimize the exploration of early patterns, which refer to initial trends or behaviors in the algorithm’s convergence process during the early stages of iterations and the exploitation of subsequent tendencies, indicating how these initial trends develop over time as the algorithm iterates and refines its search. To evaluate TASR-CHOA, we apply it to 29 conventional optimization benchmark functions, 10 IEEE CEC-06 benchmarks, 30 complicated IEEE CEC-BC benchmark functions, and ten well-known benchmark real-world challenges. We evaluate TASR-CHOA against 4 categorical optimization techniques as well as 18 top IEEE CEC-BC algorithms. Based on our broad investigation done using three statistical tests, we claim that TASR-CHOA outperforms the majority of the algorithms since within a position takes the best place, 54 out of 73 evaluation functions and engineering problems. In other cases, the results are almost the same as those of SHADE and CMA-ES over several comparisons. As an illustrative application of this joint approach, a computer-aided fire detection task is performed using a deep convolutional neural network combined with TASR-CHOA. We also outline the algorithm executed in steps, indicating computational complexity, which is O(NI×NV) + O(NI×NV×6) + O(NI×NV + 2×NI×NV) as a function of number of individuals (NI) and dimensions (NV).

Chamber shape optimization for ultra-high-pressure water-jet nozzle based on computational fluid dynamics method and a data-driven surrogate model

The ultra-high-pressure (UHP) water-jet nozzle acts as one of the key units for the whole rotary sprayers, and its chamber shape plays a decisive role in improving the hydrodynamic performance of water jetting. Unfortunately, comprehensive and effective methods to optimize the nozzles’ chamber shape are still lacking. By coupling a data-driven surrogate model with metaheuristic optimizer, a computational fluid dynamics (CFD)-based optimization scheme aiming at fully enhancing the hydrodynamic performance of UHP nozzles was proposed in this paper. Firstly, to ensure optimization accuracy, an improved whale optimization algorithm (IWOA) was developed. It combines the chaotic opposition-based learning (COBL) and the Cauchy-Gaussian mutation simulated annealing algorithms, incorporating a nonlinear convergence and adaptive inertia weight mechanism. Then, to reduce CFD computational costs and accelerate calculation speed, a data-driven surrogate model based on IWOA-support vector machine (SVM) was proposed. Herein, the parameter gamma and the penalty coefficient in SVM model, are trained using IWOA. The IWOA-SVM model was constructed using optimal Latin hypercube design (Opt. LHD) to sample nozzle structures, with peak wall shear stress (as objective function) calculated via CFD method. Finally, this optimization scheme was applied to optimize the chamber shape of the original nozzle commonly used in ship rust removal. The results reveal that the chamber shape of the optimized nozzle can greatly enhance the water-jet hydrodynamic performance by raising the peak wall shear stress by 13.88%. Additionally, the IWOA-SVM model demonstrates the capability to evaluate hydrodynamic performance with a maximum error of 2.853%. Therefore, the proposed optimization scheme provides novel insights for the overall design and optimization of the UHP water-jet nozzle.