Model Complexity Research Articles

Research on the Parameter Prediction Model for Fully Mechanized Mining Equipment Selection Based on RF-WOA-XGBoost

Fully mechanized mining equipment is core to the coal mining process. The selection process for this type of equipment is complex and heavily relies on experts’ experience for determining equipment parameters. This paper proposes a fully mechanized mining equipment parameter prediction model based on Extreme Gradient Boosting Regression Trees (XGBoost), which is developed based on the mapping relationships among geological parameters, fully mechanized mining face conditions, and the parameters of fully mechanized mining equipment. Feature selection is performed based on the feature importance ranking obtained through the Random Forest (RF) method, thereby reducing the model complexity. Different optimization algorithms are used to optimize the hyperparameters of XGBoost, and the results show that the Whale Optimization Algorithm (WOA) outperforms other algorithms in terms of convergence speed and optimization effectiveness. By comparing different prediction algorithms, it is found that the WOA-XGBoost model achieves higher prediction accuracy on the test set, with an average absolute error of 0.0458, root mean square error of 0.1610, and a coefficient of determination (R2) of 0.9451. Finally, a RF-WOA-XGBoost-based parameter prediction model for fully mechanized mining equipment is established, which is suitable for lightly inclined mining faces. This model reduces input complexity, improves the selection speed, minimizes reliance on experts, and ensures prediction accuracy, providing an effective reference for the parameter selection of fully mechanized mining equipment.

Correlation-based pruning algorithm with weight compensation for feedforward neural networks

Abstract Optimizing neural network architectures through effective pruning techniques has become essential to balancing model complexity and accuracy. This study introduces a novel correlation-based approach to systematically reduce network size by identifying and removing redundant neurons based on their activation correlations. By selectively pruning neurons while compensating for their contributions, the method maintains model fidelity across diverse datasets. Results demonstrate substantial architecture reductions with minimal performance impact: For the MNIST dataset, the number of neurons in hidden layers was reduced from 128-128 to 118-93, while maintaining a high accuracy of 97.59%. Comparative analysis indicates that this pruning approach achieves competitive or superior results compared to state-of-the-art methods while reducing computational complexity and memory requirements by up to 25%. The findings highlight the potential of correlation-driven pruning strategies to optimize neural networks, making them more efficient and adaptable to resource-constrained environments.

Penentuan Model Terbaik dalam Pengaruh Lapangan Usaha dan Nilai Ekspor Terhadap Pertumbuhan Ekonomi di Indonesia

This study aims to evaluate the effect of business sector and export value on economic growth in Indonesia using Generalised Linear Models (GLM) and Integrated Nested Laplace Approximation (INLA) approaches with Gamma and Gaussian distributions. The results showed that agriculture, forestry and fishing (PKP), mining and quarrying (PP) and fiscal (PK) sectors have a significant impact on economic growth, while real estate (RE) and export value (NE) are not significant. Based on the Akaike Information Criterion (AIC), the best model is the GLM with Gaussian distribution, which provides a balance between model complexity and the ability to explain the data. The INLA method provided consistent results with informative confidence intervals, but did not outperform the GLM in model evaluation. This study provides insights for policy makers to prioritise the PKP, PP and PK sectors in promoting economic growth and recommends further research with additional variables and more complex analytical methods for more in-depth results.

An Efficient Retinal Fluid Segmentation Network Based on Large Receptive Field Context Capture for Optical Coherence Tomography Images.

Optical Coherence Tomography (OCT) is a crucial imaging modality for diagnosing and monitoring retinal diseases. However, the accurate segmentation of fluid regions and lesions remains challenging due to noise, low contrast, and blurred edges in OCT images. Although feature modeling with wide or global receptive fields offers a feasible solution, it typically leads to significant computational overhead. To address these challenges, we propose LKMU-Lite, a lightweight U-shaped segmentation method tailored for retinal fluid segmentation. LKMU-Lite integrates a Decoupled Large Kernel Attention (DLKA) module that captures both local patterns and long-range dependencies, thereby enhancing feature representation. Additionally, it incorporates a Multi-scale Group Perception (MSGP) module that employs Dilated Convolutions with varying receptive field scales to effectively predict lesions of different shapes and sizes. Furthermore, a novel Aggregating-Shift decoder is proposed, reducing model complexity while preserving feature integrity. With only 1.02 million parameters and a computational complexity of 3.82 G FLOPs, LKMU-Lite achieves state-of-the-art performance across multiple metrics on the ICF and RETOUCH datasets, demonstrating both its efficiency and generalizability compared to existing methods.

RSM modelling and yield prediction of base-catalyzed transesterification of pumpkin-seed oil to biodiesel: artificial neural network, kinetics, and thermodynamics

The study examined the extraction of bio-oil from pumpkin seed and compared the optimization of the production of Fatty acid ethyl ester (FAEE) via the transesterification process using Response Surface Methodology (RSM), and Artificial Neural Network (ANN). This research uniquely highlights the utilization of pumpkin seed oil, a non-edible and sustainable feedstock, and combines RSM and ANN methodologies to enhance the precision of biodiesel optimization. The transesterification experiment was conducted at 60 min reaction time under varying temperature ranges, catalyst weight, stirrer speed, and ethanol-oil molar ratio. The RSM optimized conditions for maximum production were determined to be a 1.3% catalyst concentration, 6:1 ethanol-to-oil molar ratio, 50 °C temperature, and 550 rpm stirrer speed, resulting in a 90% biodiesel yield. The statistical evaluation metrics confirmed the neural network predictions were compromised to 80% yield due to limitations such as insufficient data size and the inherent complexity of the ANN model. The biofuel produced satisfied ASTM specifications peculiar to sustainable environmental applications. The mechanistic parameters revealed variations in thermodynamic stability and feasibility across reaction orders (zero, pseudo-first, and pseudo-second), emphasizing the temperature-dependent effects of the transesterification kinetic pathway on yield, design, and efficiency.

Forecasting tourist arrivals using STL-XGBoost method

Forecasting tourism demand in a timely manner is critical for ensuring the smooth operation of the tourism industry. Over time, time series models have been widely applied to estimate the number of tourists arriving. In this paper, we proposed a XGBoost model for tourism demand forecasting based on the STL seasonal decomposition. The first phase of our proposed model involves applying STL decomposition to preprocess the time series, separating it into two components: the seasonal and de-seasonal terms. During the second phase, the seasonal term is modeled and predicted with the Holt-Winters model. For the de-seasonal term, the ARIMA model is first employed to capture the residual part, Then, the XGBoost model is utilized to reconstruct both the de-seasonal term and its lag, along with the residual part obtained from the ARIMA model. By integrating the forecast outputs from both the Holt-Winters and XGBoost models, the final tourism demand predictions can be derived. The effectiveness of the proposed model is demonstrated using the tourist arrivals data in Macau from eight countries: United States, Germany, Malaysia, Philippines, India, Thailand, Italy and Korea (South Korea). The validation results indicate that the proposed model exhibits superior forecasting performance for time series data showing seasonality and trendency, simultaneously enhancing interpretability without increasing model complexity. The model outperforms five benchmark comparison models when assessed using the Symmetric Mean Absolute Percent Error (SMAPE), Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE) metrics.

Preparing MPICH for exascale

The advent of exascale supercomputers heralds a new era of scientific discovery, yet it introduces significant architectural challenges that must be overcome for MPI applications to fully exploit its potential. Among these challenges is the adoption of heterogeneous architectures, particularly the integration of GPUs to accelerate computation. Additionally, the complexity of multithreaded programming models has also become a critical factor in achieving performance at scale. The efficient utilization of hardware acceleration for communication, provided by modern NICs, is also essential for achieving low latency and high throughput communication in such complex systems. In response to these challenges, the MPICH library, a high-performance and widely used Message Passing Interface (MPI) implementation, has undergone significant enhancements. This paper presents four major contributions that prepare MPICH for the exascale transition. First, we describe a lightweight communication stack that leverages the advanced features of modern NICs to maximize hardware acceleration. Second, our work showcases a highly scalable multithreaded communication model that addresses the complexities of concurrent environments. Third, we introduce GPU-aware communication capabilities that optimize data movement in GPU-integrated systems. Finally, we present a new datatype engine aimed at accelerating the use of MPI derived datatypes on GPUs. These improvements in the MPICH library not only address the immediate needs of exascale computing architectures but also set a foundation for exploiting future innovations in high-performance computing. By embracing these new designs and approaches, MPICH-derived libraries from HPE Cray and Intel were able to achieve real exascale performance on OLCF Frontier and ALCF Aurora respectively.

An ensemble multi-dimensional randomization network for intelligent recognition of tobacco baking stage

In recent years, image processing technology has been increasingly studied on intelligent unmanned platforms, and the differences in the shooting environment during tobacco baking pose challenges to image processing algorithms. To address this problem, an ensemble multi-dimensional randomization network (EMRNet) for intelligent recognition of tobacco baking stage is proposed. The first is to obtain the tobacco leaf area during the baking process. Then, a multi-dimensional randomization network (MRNet) is designed to recognize tobacco baking stage. The effectiveness of MRNet lies in multi-scale hidden layer feature extraction, which can effectively enhance the expression ability of features to overcome the impact of differences between different environments on the tobacco baking stage. Finally, MRNet is used as component learner for constructing an ensemble randomization network structure to distinguish the tobacco baking stage. On the constructed tobacco baking stage dataset, EMRNet achieves 89.14% accuracy with 642.96MFLOPs. Compared with SVM, MLP, BP, ELM, CRVFL and other algorithms, EMRNet shows excellent performance in accuracy and model complexity. The proposed method explores the application of image processing technology in crop baking and drying, providing theoretical support for intelligent baking technology.

Accurate colorectal cancer detection using a random hinge exponential distribution coupled attention network on pathological images.

Colorectal cancer (CRC) is one of the most common and deadly forms of cancer worldwide, necessitating accurate and early detection to improve treatment outcomes. Traditional diagnostic methods often rely on manual examination of pathological images, which can be time-consuming and prone to human error. This study presents an advanced approach for colorectal cancer detection using a Random Hinge Exponential Distribution coupled Attention Network (RHED-CANet) on pathological images. The input dataset is sourced from the TCGA-CRC-DX cohort and the CRC dataset, both widely recognized for their comprehensive coverage of colorectal cancer cases. Pre-processing and feature extraction are performed using a Modified Square Root Sage-Husa Adaptive Kalman Filter combined with a Spike-Driven Transformer, enhancing noise reduction and feature clarity. Segmentation is achieved through an EfficientNetV2L Inception Transformer, ensuring precise delineation of cancerous regions. The final classification utilizes the RHED-CANet, a network tailored to handle the complexities of pathological data with high accuracy. This methodology achieved remarkable results, with an accuracy of 99.9% and a precision of 99.7%. These performance metrics underscore the method's ability to minimize false positives and enhance diagnostic accuracy. The proposed approach offers significant advantages, including a reduction in diagnostic time and a substantial improvement in detection accuracy, making it a promising tool for clinical applications. Despite its excellent accuracy, the suggested RHED-CANet technique has drawbacks, such as overfitting the TCGA-CRC-DX and CRC datasets by reducing generalizability on other datasets comprising other cancer types or image qualities. The actual application of the techniques in real-time clinical applications may be hampered by this computational load, especially in settings with limited resources, and the model's potential computational complexity due to multiple advanced processing steps. Additionally, the efficiency of training may be impacted by biased inputs, particularly for minor CRC subtypes.

Performance of Classification for Adjusted Adaptive Elastic Net Penalty and Adaptive LASSO for Breast Cancer Data

In The current time witnesses a significant surge in data, fueled by technology's rapid advancement. This increase in data volume has led to the emergence of high-dimensional data (where the number of variables exceeds the sample size), creating challenges in precision and target identification. Consequently, binary response variable classification becomes intricate due to the multicollinearity in explanatory variables. To tackle this, response variable classification has prompted the utilization of penalization techniques, reduced variables and selecting best variables in the model. This aids in simplifying the model complexity to attain the specific binary outcome (0,1). In this paper, penalization methods were applied, including Adaptive Least Absolute Shrinkage and Selection Operator, With the Adjusted Adaptive Elastic Net Penalty with logistic regression model. The application involved a set of real data. A sample collected by the researcher (p=49, n=41), and it produced positive results for classification in the sample collected by the researcher, in resulted we found these methods have achieving high classification accuracy while efficiently selecting an optimal number of variables using a range of packages and functions in the R programming language.

Improving functional protein generation via foundation model-derived latent space likelihood optimization.

A variety of deep generative models have been adopted to perform de novo functional protein generation. Compared to 3D protein design, sequence-based generation methods, which aim to generate amino acid sequences with desired functions, remain a major approach for functional protein generation due to the abundance and quality of protein sequence data, as well as the relatively low modeling complexity for training. Although these models are typically trained to match protein sequences from the training data, exact matching of every amino acid is not always essential. Certain amino acid changes (e.g., mismatches, insertions, and deletions) may not necessarily lead to functional changes. This suggests that maximizing the training data likelihood beyond the amino acid sequence space could yield better generative models. Pre-trained protein large language models (PLMs) like ESM2 can encode protein sequences into a latent space, potentially serving as functional validators. We propose training functional protein sequence generative models by simultaneously optimizing the likelihood of training data in both the amino acid sequence space and the latent space derived from a PLM. This training scheme can also be viewed as a knowledge distillation approach that dynamically re-weights samples during training. We applied our method to train GPT- like models (i.e., autoregressive transformers) for antimicrobial peptide (AMP) and malate dehydrogenase (MDH) generation tasks. Computational experiments confirmed that our method outperformed various deep generative models (e.g., generative adversarial net, variational autoencoder, and GPT model without the proposed training strategy) on these tasks, demonstrating the effectiveness of our multi-likelihood optimization strategy.

Estimating evolutionary and demographic parameters via ARG-derived IBD.

Inference of evolutionary and demographic parameters from a sample of genome sequences often proceeds by first inferring identical-by-descent (IBD) genome segments. By exploiting efficient data encoding based on the ancestral recombination graph (ARG), we obtain three major advantages over current approaches: (i) no need to impose a length threshold on IBD segments, (ii) IBD can be defined without the hard-to-verify requirement of no recombination, and (iii) computation time can be reduced with little loss of statistical efficiency using only the IBD segments from a set of sequence pairs that scales linearly with sample size. We first demonstrate powerful inferences when true IBD information is available from simulated data. For IBD inferred from real data, we propose an approximate Bayesian computation inference algorithm and use it to show that even poorly-inferred short IBD segments can improve estimation. Our mutation-rate estimator achieves precision similar to a previously-published method despite a 4 000-fold reduction in data used for inference, and we identify significant differences between human populations. Computational cost limits model complexity in our approach, but we are able to incorporate unknown nuisance parameters and model misspecification, still finding improved parameter inference.

Bayesian deep learning applied to diabetic retinopathy with uncertainty quantification.

Deep Learning (DL) has significantly contributed to the field of medical imaging in recent years, leading to advancements in disease diagnosis and treatment. In the case of Diabetic Retinopathy (DR), DL models have shown high efficacy in tasks such as classification, segmentation, detection, and prediction. However, DL model's opacity and complexity lead to errors in decision-making, particularly in complex cases, making it necessary to estimate the model's uncertainty in predictions. Therefore, there is a need to estimate uncertainty in the model's predictions, which cannot be estimated by classical DL models alone. To address this issue, Bayesian DL methods have been proposed, and their use is increasing in the field. In this paper, we developed a straightforward architecture for the classification of DR using a Convolutional Neural Network (CNN) model. We then applied the Bayesian CNN twice, once using Variational Inference (VI) and once using Monte Carlo dropout (MC-dropout) methods, to the same CNN architecture. This allowed us to gain the posterior predictive distributions for each of them. The performance of the proposed models was evaluated on two benchmark datasets, namely APTOS 2019 and Messidor-2. Experimental findings demonstrated that the proposed models surpassed other state-of-the-art models, achieving a test accuracy of 94.70% and 77.00% for CNN, 94.00% and 86.00% for BCNN-VI, and 93.30% and 81.00% for BCNN-MC-dropout on the APTOS dataset and Messidor-2 dataset, respectively. Finally, we computed the entropy and standard deviation on the obtained predictive distribution to quantify the model uncertainty. This research highlights the potential benefits of using Bayesian DL methods in medical image analysis to improve the accuracy and reliability of diagnosing disease and treatment.

A deep neural network-based multi-layer classifier ensembles for intrusion detection in fog-based Internet of Things environments

Intrusion Detection Systems (IDS) are regarded as an efficient security mechanism in the Internet of Things (IoT). IDSs in the IoT face critical challenges, such as a vast amount of data with many features, limited resources, detection of unknown attacks, and many false alarms. To overcome these issues, this paper proposes a novel framework, namely a Multi-Layer Multi-Classification System (MLMCS). The proposed framework consists of four steps: In the first step, the data collected from information sources are pre-processed. The second step divides the feature space into several subspaces to reduce the model's complexity. The third step proposes a multilayer base classifier called a Multi-Layer Classifier System (MLCS) for each feature subspace. Each MLCS is layered to transform a multi-class classification into several binary classifications so that a specific category of attacks has been detected in each layer based on the Group Method of Data Handling (GMDH) neural network. In the fourth step, the results obtained from MLCSs based on the Dempster-Shafer theory are combined to reduce the problems related to overfitting. The simulation results were gained using three datasets, i.e., NSL-KDD, UNSW-NB15, and TON-IoT. They indicated that the proposed method, compared with other methods, improved F1-score by 5.64%, 0.65%, and 0.11%, respectively. Furthermore, the false alarm rate was reduced by 2.66%, 2.52%, and 0.02%. Moreover, the impact of four well-known adversarial attacks on the dataset was investigated. It was observed that the MLMCS method, with an average 5.21% reduction, could perform well against these attacks.

A novel hybrid deep learning model for ultra-short-term prediction of wind speed

Accurate ultra-short-term wind speed prediction is critical for various engineering applications. Although decomposition-integration methods are widely used for this purpose, several challenges remain. This study addresses these challenges by integrating two-stage feature extraction, a combination weight prediction model, and intelligent integration to improve prediction accuracy. In the decomposition phase, two advanced methods are employed to reduce data complexity and extract comprehensive features. During reconstruction, multiscale sample entropy is applied to balance computational efficiency with model complexity. To overcome the limitations of individual forecasting models, a combined prediction model incorporating two deep learning approaches is developed, with weights adaptively optimized using the Sparrow Search Algorithm. Additionally, to address the variability in subsequence contributions, intelligent weight integration based on deep learning models is implemented, ensuring robust and accurate final predictions. Validation using wind speed data from three Hong Kong Observatory weather stations demonstrates that the proposed method outperforms benchmark models, achieving more-satisfactory accuracy, stability, and robustness.

Segmentation of Brain Tumor Resections in Intraoperative 3D Ultrasound Images Using a Semisupervised Cross nnSU‐Net

ABSTRACTIntraoperative ultrasound (iUS) has been widely used in recent years to track intraoperative brain tissue deformation. Outlining tumor boundaries on iUS not only facilitates the robustness and accuracy of brain shift correction but also enables the direct use of iUS information for neurosurgical navigation. We developed a semisupervised cross nnU‐Net with depthwise separable convolution (SSC nnSU‐Net) for real‐time segmentation of 3D iUS images by two networks with different initialization but consistent network structure networks. Unlike previous methods, RESECT as labeled data and ReMIND as unlabeled data for hybrid dataset training selected break down the barriers between different datasets and further alleviate the problem of “data hunger.” The SSC nnSU‐Net method was evaluated by ablation of semisupervised learning, comparison with other state‐of‐the‐art methods, and model complexity. The results indicate that the proposed framework achieves a certain balance in terms of computation time, GPU memory utilization, and segmentation performance. This motivates segmentation of 3D iUS images for real‐time application in clinical surgery. The method can assist surgeons in identifying brain tumors through iUS.

MetaRange.jl: A Dynamic and Metabolic Species Range Model for Plant Species.

Process-based models for range dynamics are urgently needed due to increasing intensity of human-induced biodiversity change. Despite a few existing models that focus on demographic processes, their use remains limited compared to the widespread application of correlative approaches. This slow adoption is largely due to the challenges in calibrating biological parameters and the high computational demands for large-scale applications. Moreover, increasing the number of simulated processes (i.e., mechanistic complexity) may further exacerbate those reasons of delay. Therefore, balancing mechanistic complexity and computational effectiveness of process-based models is a key area for improvement. A promising research direction is to expand demographically explicit metapopulation models by integrating metabolic constraints. We translated and expanded a previously developed R metapopulation model to Julia language and published it as a Julia module. The model integrates species-specific parameters such as preferred environmental conditions, biomass and dispersal ability with demographic rates (e.g., reproductive and mortality rates) derived from local temperature and biomass via the metabolic theory of ecology. We provide a simple application example for the model in which we illustrate a typical use case by predicting the future occurrence of Orchis militaris in Bavaria under different climate change scenarios. Our results show that climate change reduces habitat suitability overall, but some regions like the Franconian Forest and the Alps see increased suitability and abundance, confirming their role as refugia. Simulating metapopulation dynamics reveals that local population dynamics and dispersal are crucial for accurate predictions. For instance, increasing dispersal distance reduces overall abundance loss but also lessens population growth in refugia. This highlights the importance of measuring traits like dispersal ability to improve climate change forecasts.

Radionuclide transport and retardation in uplifting granitic rocks: Part 2 – Modelling coupled processes in uplift scenarios

Uplifting fractured granitic rocks occur in substantial areas of countries such as Japan. Some of these areas might be considered when siting a deep geological repository for radioactive wastes. A repository site would be selected in such an area only if it is possible to make a safety case, accounting for the changing conditions during uplift. The safety case must include robust arguments that chemical processes in the rocks around the repository will contribute sufficiently to minimise radiological doses to biosphere receptors. Numerical modelling is an important aspect of making these arguments. To provide confidence in the safety arguments, numerical models need to be sufficiently realistic, but also parameterised conservatively (pessimistically). However, model development is challenging because uplift involves many complex couplings between groundwater flow, chemical reactions between water and rock, and changing rock properties. The couplings would affect radionuclide mobilisation and retardation, by influencing diffusive radionuclide fluxes between groundwater flowing in fractures and effectively immobile porewater in the rock matrix (rock matrix diffusion, RMD) and radionuclide partitioning between water and solid phases, via: (i) mineral precipitation/dissolution; (ii) mineral alteration; and (iii) sorption/desorption. It is difficult to represent all this complexity in numerical models while showing that they are parameterised conservatively. Here we present a modelling approach, illustrated by simulation cases for some exemplar radioelements, to identify realistically conservative process conceptualisations and model parameterisations.

Advancing river flood forecasting with a collaborative integrated modeling method.

River flood forecasting and assessment are crucial for reducing flood risks, as they offer early alerts and allow for proactive actions to safeguard individuals from possible flood-related damage. Effective modeling in this field often multiple interconnected aspects of the hydrologic cycle, such as precipitation, infiltration, runoff, and evaporation, requiring collaboration among hydrology experts. Such collaboration enables experts to handle and manage their specialized processes more effectively, thereby enhancing the efficiency of the development of integrated flood forecasting models. Tight integration and loose integration are two common strategies for integrating different hydrologic cycle process models in river flood forecasting. However, most integration strategies rely on centralized models, necessitating experts to configure models and data on local computers. Currently, there is a deficiency in the capacity for effective collaboration in the integrated modeling of river flood forecasts. This issue arises from multiple obstacles: the complexity of understanding heterogeneous data and hydrologic cycle process models; the difficulty of integrating models with diverse runtime environments; and the challenge of synchronizing forecasting model changes among experts in real time. Therefore, we propose a web-based collaborative integrated modeling method, designed to support both tightly and loosely integrated modes, to enhance collaborative river flood forecasting and assessment. This method includes three core modules: (1) data and model description for providing a structured description of the execution logic of forecasting models and the internal structure of forecast data for expert understanding; (2) model access and integration for access and integration of data and multi-source heterogeneous models of hydrologic cycle processes; and (3) modeling scenario configuration for collaborative development of forecasting models and the execution of simulation tasks. Finally, we illustrate the application of the proposed method by utilizing the GEFS v12 (Global Ensemble Forecast System) rainfall ensemble forecasting dataset with the CREST (Coupled Routing and Excess STorage) hydrologic model. The results show enhanced efficiency in the collaborative development of river flood forecasts by hydrology experts, particularly in model accessibility, data processing, simulation, and evaluation, thereby potentially aiding decision-making.

Aircraft Load Estimation Using Linear Parameter-Varying System-Based Hybrid Observers

The hybrid loads observer has demonstrated its precision in estimating structural loads and wind disturbances in prior applications. This level of precision is attained by combining a high-fidelity physical, nonlinear flight dynamics model with a data-driven correction model. To mitigate the typically high computational effort, the nonlinear model is substituted in this work with a linear parameter-varying (LPV) system. Therefore, the nonlinear model is approximated by scheduled Linear Time-Invariant models derived from Jacobian linearization. The robustness of this novel approach against parameter uncertainties is evaluated through simulated flight test studies using a subscale test aircraft as an example. It is demonstrated that increased model uncertainties lead to wind estimation accuracy reduction. Additionally, increased parameter uncertainties adversely affect the accuracy of structural load estimation within the model-based (physical) part of the observer. Nonetheless, this loss of accuracy can be compensated by the correction model, leading to the typical high load estimation accuracy of the hybrid loads observer. Finally, experimental data from wind-tunnel tests, utilizing a 1-degree-of-freedom representative test wing, confirms the high estimation accuracy of the LPV-based hybrid loads observer. Despite employing low-fidelity models, achieving high accuracy is feasible while maintaining the characteristic low complexity of the correction model.