Adaptive Model Research Articles

Adaptive Bi-Encoder Model Selection and Ensemble for Text Classification

Can bi-encoders, without additional fine-tuning, achieve a performance comparable to fine-tuned BERT models in classification tasks? To answer this question, we present a simple yet effective approach to text classification using bi-encoders without the need for fine-tuning. Our main observation is that state-of-the-art bi-encoders exhibit varying performance across different datasets. Therefore, our proposed approaches involve preparing multiple bi-encoders and, when a new dataset is provided, selecting and ensembling the most appropriate ones based on the dataset. Experimental results show that, for text classification tasks on subsets of the AG News, SMS Spam Collection, Stanford Sentiment Treebank v2, and TREC Question Classification datasets, the proposed approaches achieve performance comparable to fine-tuned BERT-Base, DistilBERT-Base, ALBERT-Base, and RoBERTa-Base. For instance, using the well-known bi-encoder model all-MiniLM-L12-v2 without additional optimization resulted in an average accuracy of 77.84%. This improved to 89.49% through the application of the proposed adaptive selection and ensemble techniques, and further increased to 91.96% when combined with the RoBERTa-Base model. We believe that this approach will be particularly useful in fields such as K-12 AI programming education, where pre-trained models are applied to small datasets without fine-tuning.

Facilitators and Inhibitors to the Application of the Acuity Adaptable Model.

Facilitators and Inhibitors to the Application of the Acuity Adaptable Model.

An adaptive dynamical system model for development of schizophrenia: Epigenetics and false memories

This paper introduces a fifth-order adaptive dynamical system model represented as a self-modelling network model to understand the role of epigenetic factors in developing schizophrenia, particularly about the occurrence of false memories. Schizophrenia, affecting a significant portion of the global population, presents complex challenges due to its symptoms across cognitive, emotional, and social domains. Central to this paper is the exploration of epigenetic influences on the development of schizophrenia, with a focus on how these epigenetic factors contribute to the generation of false memories, a notable symptom of the disorder. The model assumes that early adverse experiences significantly affect hippocampal development, which is crucial for memory processes. By employing a fifth-order adaptive dynamical system model represented as a self-modelling network, this paper examines the intricate dynamics between genetic predispositions, environmental factors, and epigenetic modifications in schizophrenia. The model’s sophistication allows for a nuanced understanding of the gene-environment interactions, particularly highlighting the role of early-life trauma in shaping lifelong cognitive outcomes. This approach offers a comprehensive framework for analyzing the multifaceted nature of schizophrenia, contributing to a deeper understanding of its etiology and progression.

An adaptable Digital Twin model for manufacturing

In a smart manufacturing system, physical assets are often connected to their Digital Twins in the virtual world that composes a cyber-physical production system (CPPS). Digital Twin is synchronized with the physical asset. Digital Twin is therefore considered an enabling technology of smart manufacturing. In the process of developing Digital Twins, the Digital Twin model is of great significance. Many Digital Twin models have been proposed in the past decades, including geometry models, information models, and ontology models. However, there has not been an adaptable and generic Digital Twin model that can be used for various scenarios in a manufacturing system. This paper proposes an adaptable Digital Twin model based on the object-oriented concept. The proposed Digital Twin model can be applied to different situations using object-oriented concept features. Firstly, the base Digital Twin model is developed according to ISO 23247, including the essential attributes of a Digital Twin. The aggregation and composition mechanism of the proposed Digital Twin model is also described based on the object-oriented concept. After that, according to the categories of the physical assets on the shop floor, Digital Twin models of different physical assets are developed based on the base Digital Twin model, inheriting the features of the base Digital Twin model. In the end, a case study is carried out for verification on a machine tool in the Laboratory for Industry 4.0 Smart Manufacturing Systems (LISMS).

A novel dynamic structural adaptive multivariable grey model and its application in China's solar energy generation forecasting

Accurately predicting China's solar energy generation is crucial for energy planning and environmental protection. However, the nonlinearity and complexity of the data pose challenges to prediction. This article proposes a dynamic structural adaptive multivariate grey model to overcome these challenges. The model constructs a dynamic nonlinear correction function with an uncertain structure that can adaptively adjust to the data characteristics. Additionally, the Gaussian function is utilized to derive the time response function, overcoming the mechanism defects present in traditional multivariate grey models. The whale optimization algorithm is applied to optimize the background value of the model, improving accuracy. The model is compatible with 14 classical models via parameter changes and has strong compatibility. The results show that the fitting error and test error of the model are 0.23 % and 1.95 % respectively, which is significantly better than the other 11 methods, indicative of strong robustness and accuracy. The forecast indicates that China's solar energy generation will experience a consistent upward trajectory from 2023 to 2030, with an average annual growth rate of 12.11 %. This study significantly contributes to the advancement of the theory of multivariate grey dynamic prediction and provides essential data support for strategic planning within China's solar power industry.

Novel flow field design for redox flow battery using algorithmic channel generation and self-adaptive network model and experimental verification

Flow fields play a vital role in redox flow batteries (RFBs) to increase the power density. Traditional design for flow field usually employs CFD simulations with substantial computational costs and adopts enumeration and trial-and-error methods based on designers’ subjectivity, which have become a bottleneck in flow field design. In this study, a novel research paradigm is employed for flow field design in vanadium redox flow battery (VRFB), i.e., an adaptive three-dimensional equivalent network model is developed and a self-developed flow channel generation algorithm is employed to obtain optimum flow field through comparison and screening. Firstly, a flow channel library consisting of 18,135 flow fields is generated by self-developed flow channel generation algorithm. Secondly, the electrolyte flow, species transport and charge transport with various flow fields is researched by the self-adaptive equivalent network model with low computational costs due to eliminated meshing. Thirdly, pressure drop and voltage efficiency are employed to evaluate different flow fields for the aim of obtaining high-performance flow field structures through comparison and screening. The obtained high-performance flow field is fabricated and its high performance is confirmed by comprehensive experimental validation. The energy efficiency of the high-performance flow field is 4 % higher than that of the serpentine flow field investigated at 100 mA cm−2. This work employs adaptive equivalent network model to generate and optimize flow channel for the first time, which provides a reference for flow channel selection and design.

Adaptive morphological modeling and optimal estimation for dynamic LVET measurement from seismocardiographic signals

The dynamic left ventricular ejection time (LVET) represents a crucial cardiovascular health indicator. Seismocardiography (SCG) offers substantial advantages over conventional techniques, however, it is susceptible to motion artifacts and exhibits complex morphologies, posing challenges for accurate dynamic LVET measurement. This study proposes a synergistic approach that integrates adaptive morphological modeling with an optimal estimation technique to accurately measure dynamic LVET from SCG signals. Firstly, an active feature recognition method was proposed to automatically identify 10 morphological features, followed by modeling them using 10 Gaussian functions. Subsequently, the model results were fused with measured SCG through optimal estimation, involving adaptive updating of the measurement noise matrix. When evaluated against state-of-the-art algorithms on the open-source CEBS dataset (82997 heartbeats) and the experimental dataset (34181 heartbeats), the proposed algorithm exhibits outstanding performance, with a goodness of fit (R2) of 0.9272 and root-mean-square error (RMSE) of 13.808 ms, convincingly demonstrating its effectiveness and advancement.

Real-time adaptive model of mainstream parameters for aircraft engines based on OSELM-EKF

Gas path parameters play a crucial role in the health management of aeroengines, which are usually generated from the nonlinear component-level model. However, risks of stability and substantial time consumption confine online applications of conventional models. This paper proposes a mainstream parameter model that exclusively involves rotating components using fewer gas path parameters, including an adaptive strategy based on an online sequential extreme learning machine and an extended Kalman filter. The mainstream parameter model is designed in a linear parameter-varying form with a non-iterative smooth state-switching mechanism and enables real-time operation by simplifying the component complexity. Besides, some sensor measurements are employed to update rotor speeds, thus eliminating the need for derivative computations. Neural networks are introduced in compressor component calculations. Additionally, the extended Kalman filter is developed to estimate health parameters to tune the system equation residuals, and the learning machine is applied to compensate for rotating components’ pressure ratios under different degradation magnitudes. Finally, systematic tests are carried out to evaluate the computation accuracy and fast capabilities of the mainstream parameter adaptive model in various scenarios. Simulations demonstrate the proposed methodology's superiority over traditional adaptive correction schemes.

MERLIN, an adaptative LTE radiative transfer model for any mixture: Validation on Eurofer97 in argon atmosphere

Optical Emission Spectroscopy is used within the LIBS (Laser-Induced Breakdown Spectroscopy) technique to measure the elemental composition of a sample irradiated by a laser pulse. When the objective is to characterize a known alloy or to analyze the sample semi-quantitatively, standards can be used. This method refers to the “calibrated LIBS”. When the studied sample is complex (for instance unknown alloy and/or concentration gradients), the iterative modeling of the experimental emission spectra leads to the determination of the laser-induced plasma characteristics including its composition. This second method refers to the “calibration-free LIBS”.In this paper, we present the radiative transfer code MERLIN (MultiElemental Radiative equiLibrIum emissioN) aiming at matching experimental spectra of a laser-induced plasma in Local Thermodynamic Equilibrium (LTE) conditions for calibration-free LIBS. MERLIN provides the simulated spectra for any elementary mixture thanks to queries from spectral databases available online. The code is optimized and modular to allow outputs even for a high number of species.The validation of MERLIN presented in the paper is based on a thorough analysis of the plasma emission performed on the alloy Eurofer97 in experimental conditions providing the LTE. The reconstruction of the observed spectra is performed with MERLIN. For the reconstruction, no adjusted variable is required since all the necessary parameters are derived from experiments.

An Integrated Imitation and Reinforcement Learning Methodology for Robust Agile Aircraft Control with Limited Pilot Demonstration Data

In this paper, we present a methodology for constructing data-driven maneuver generation models for agile aircraft that can generalize across a wide range of trim conditions and aircraft model parameters. Maneuver generation models play a crucial role in the testing and evaluation of aircraft prototypes, providing insights into the maneuverability and agility of the aircraft. However, constructing the models typically requires extensive amounts of real pilot data, which can be time-consuming and costly to obtain. Moreover, models built with limited data often struggle to generalize beyond the specific flight conditions covered in the original dataset. To address these challenges, we propose a hybrid architecture that leverages a simulation model, referred to as the source model. This open-source agile aircraft simulator shares similar dynamics with the target aircraft and allows us to generate unlimited data for building a proxy maneuver generation model. We then fine-tune maneuver generation model to the target aircraft using a limited amount of real pilot data. Our approach combines techniques from imitation learning, transfer learning, and reinforcement learning to achieve this objective. To validate our methodology, we utilize real agile pilot data provided by Turkish Aerospace Industries (TAI). By employing the F-16 as the source model, we demonstrate that it is possible to construct a maneuver generation model that generalizes across various trim conditions and aircraft parameters without requiring any additional real pilot data. Our results showcase the effectiveness of our approach in developing robust and adaptable maneuver generation models for agile aircraft.

A cooperative lane change control strategy for cooperative adaptive cruise control platoons with insufficient headway gaps

Cooperative Adaptive Cruise Control (CACC) platoons contribute to enhancing road traffic safety and efficiency. However, creating suitable headway gaps for CACC platoons to change lanes is a challenge. This paper proposes a cooperative lane change control strategy for CACC platoons in mixed traffic environments where suitable headway gaps on the target lane are insufficient. Firstly, a cooperative longitudinal and lateral control strategy for CACC platoons is designed. Once longitudinal control ensures that the headway gap on the target lane meets the criteria, the vehicles are controlled to complete the lane change and resume cruising in their original formation. To achieve the longitudinal control objectives of gap generation by cooperative deceleration before lane change and cruising after lane change, a longitudinal cooperative control strategy that can independently create the lane change gap is proposed based on an adaptive weight model predictive control. Finally, the lane change headway gaps in the proposed strategy are generated sequentially for each vehicle, a contrast experiment is designed where the lane change headway gap is generated once and vehicles change lanes simultaneously. The effectiveness of the proposed strategy is validated through numerical simulation combining CarSim and Matlab/Simulink. The results show that the CACC platoon completed the lane change successfully under the control of the proposed strategy. Compared to the contrast simulation, in both low-speed and high-speed highway scenarios, the average speed standard deviation under the proposed strategy is reduced by 48.3 % and 39.9 % respectively, while the average comfort index decreases by 11.7 % and 5.9 %.

Adaptive multi-surrogate model for complex opto-mechanical system alignment

Owing to the inherent nonlinearity of surrogate models, they have been widely used in the alignment of complex opto-mechanical systems. Finding the most suitable surrogate model for misalignment wavefront errors is challenging. This study proposes an adaptive multi-surrogate model (AMSM) based on the concept of layer-by-layer approximation and adaptive selection and a compensation mechanism to facilitate the combination of the AMSM and non-dominated sorting genetic algorithm II (NSGA-II). For example, alignment simulations of an off-axis three-mirror anastigmatic (TMA) telescope and a catadioptric microscope objective were performed. The results show that the AMSM is well done in terms of approximate accuracy and efficiency. The application of the proposed method in the off-axis TMA telescope leads to 65.2% and 18.7% reductions in the full-field average RMS wavefront error compared with the traditional sensitivity table method (STM) and radial basis function (RBF) network surrogate model, respectively. Similarly, for the catadioptric microscope objective, there is a reduction of 43.5% and 22.4%, respectively. The AMSM will further promote the development of a surrogate model to align more complex and precise opto-mechanical systems.

A method framework of semi-automatic knee bone segmentation and reconstruction from computed tomography (CT) images.

Accurate delineation of knee bone boundaries is crucial for computer-aided diagnosis (CAD) and effective treatment planning in knee diseases. Current methods often struggle with precise segmentation due to the knee joint's complexity, which includes intricate bone structures and overlapping soft tissues. These challenges are further complicated by variations in patient anatomy and image quality, highlighting the need for improved techniques. This paper presents a novel semi-automatic segmentation method for extracting knee bones from sequential computed tomography (CT) images. Our approach integrates the fuzzy C-means (FCM) algorithm with an adaptive region-based active contour model (ACM). Initially, the FCM algorithm assigns membership degrees to each voxel, distinguishing bone regions from surrounding soft tissues based on their likelihood of belonging to specific bone regions. Subsequently, the adaptive region-based ACM utilizes these membership degrees to guide the contour evolution and refine segmentation boundaries. To ensure clinical applicability, we further enhance our method using the marching cubes algorithm to reconstruct a three-dimensional (3D) model. We evaluated the method on six randomly selected knee joints. We evaluated the method using quantitative metrics such as the Dice coefficient, sensitivity, specificity, and geometrical assessment. Our method achieved high Dice scores for the femur (98.95%), tibia (98.10%), and patella (97.14%), demonstrating superior accuracy. Remarkably low root mean square distance (RSD) values were obtained for the tibia and femur (0.5±0.14 mm) and patella (0.6±0.13 mm), indicating precise segmentation. The proposed method offers significant advancements in CAD systems for knee pathologies. Our approach demonstrates superior performance in achieving precise and accurate segmentation of knee bones, providing valuable insights for anatomical analysis, surgical planning, and patient-specific prostheses.

Adaptive Tube-Based Model Predictive Control for the Receiving-Side DC–DC of Dynamic Wireless Power Transfer System

Adaptive Tube-Based Model Predictive Control for the Receiving-Side DC–DC of Dynamic Wireless Power Transfer System

Unsupervised detection and mapping of sparks in the Electrochemical Discharge Machining (ECDM) process

Material removal in electrochemical discharge machining is caused by sparks generated in a tool immersed in an electrolytic solution. Being the primary machining agent in this non-contact machining process, mapping the locations of microscopic sparks is of great interest. The distribution of sparks around the tool surface could give insights into the machined hole properties like the size, surface finish, and depth as compared to the machining parameters such as applied voltage, tool size, rotation speed, and feed rate. This paper is focused on detecting sparks in photographs of the ECDM process captured using a high-speed camera. A novel approach of using a tri-planar reflective surface for capturing the location of sparks in 3D space using a 2D camera output is attempted. Traditional spark detection methods use neural network classifiers that need labeled data for training. This labeled data often comes from human intervention and contains inherent biases that could lead to misclassification. In this paper, an unsupervised spark detection methodology is demonstrated, which eliminates the need for human intervention and relies on the number of neighboring pixels detected in regions of interest (ROIs). The feasibility of using adaptive background modeling to classify thousands of images and identify the ones with sparks is demonstrated in this work. The masking technique combining effects of erosion followed by dilation is used to determine the exact boundaries of the spark contours in every image. Centroids for each of these contours are then transformed from the skewed coordinate system as observed in camera images, to a three-dimensional orthogonal coordinates system centered around the tool. The same procedure is repeated for various voltages to benchmark the distribution of sparks around a tool tip in an ECDM process.

Fostering TBL Success at Alfaisal University: A Complex Adaptive Systems Approach.

This study explores the implementation of Team-Based Learning (TBL) at Alfaisal University's College of Medicine through the lens of Complex Adaptive Systems (CAS) theory. The research question investigates how the application of CAS principles can enhance the implementation and effectiveness of TBL in medical education. The study employed a convergent parallel mixed methods longitudinal design, integrating quantitative performance metrics and qualitative themes. Quantitative analysis revealed modest improvements in individual and team-based learning scores, with a promising trend of students moving from the lower to the higher quartiles over time. Qualitative insights aligned with CAS principles, highlighting the adaptive implementation, emergent outcomes, self-organization, positive feedback loops, and depth of learning facilitated by TBL. The findings demonstrate the value of a CAS-informed approach in navigating the complexities of educational change and fostering a more resilient and adaptive educational model. The study contributes to the understanding of how CAS theory can guide the successful implementation of innovative pedagogies like TBL in medical education.

Optimal hydrogen-battery energy storage system operation in microgrid with zero-carbon emission

To meet the greenhouse gas reduction targets and address the uncertainty introduced by the surging penetration of stochastic renewable energy sources, energy storage systems are being deployed in microgrids. Relying solely on short-term uncertainty forecasts can result in substantial costs when making dispatch decisions for a storage system over an entire day. To mitigate this challenge, an adaptive robust optimization approach tailored for a hybrid hydrogen battery energy storage system (HBESS) operating within a microgrid is proposed, with a focus on efficient state-of-charge (SoC) planning to minimize microgrid expenses. The SoC ranges of the battery energy storage (BES) are determined in the day- ahead stage. Concurrently, the power generated by fuel cells and consumed by electrolysis device are optimized. This is followed by the intraday stage, where BES dispatch decisions are made within a predetermined SoC range to accommodate the uncertainties realized. To address this uncertainty and solve the adaptive optimization problem with integer recourse variables in the intraday stage, we proposed an outer-inner column-and-constraint generation algorithm (outer-inner-CCG). Numerical analyses underscored the high effectiveness and efficiency of the proposed adaptive robust operation model in making decisions for HBESS dispatch.

A two-stage adaptive consensus reaching process with improved automatic strategy for multi-attribute large group emergency decision-making

The multi-attribute large group emergency decision-making (MALGEDM) problem has gained increasing attention and becomes an important topic in the field of decision-making. The existing consensus models, however, rarely consider the effective balance between experts’ willingness and time consumption. To address this problem, we design a two-stage adaptive consensus reaching process that combines two models of the minimum adjustment amount and the minimum iteration number in the MALGEDM context. The process can achieve consensus with an improved automatic strategy that considers experts’ willingness. Firstly, the self-organizing map algorithm is employed to classify large group heterogeneous experts into multiple subgroups for the dual decision of clustering and expert categories. Secondly, we classify different degrees of consensus into four scenarios (low, medium, medium-high, and high). On this basis, we have constructed a two-stage adaptive consensus model that incorporates the novel possibility degree. The minimum adjustment amount of the adaptive feedback process can be obtained by the optimization model in the first stage, which is used as a constraint for considering the experts’ willingness in the second stage. Finally, the practicality and timeliness of the proposed model is verified through an illustrative example of emergency decision-making. Moreover, comparative analyses are conducted to highlight the superiority of the proposed model.

Self-triggered MPC with adaptive prediction horizon for nano-satellite attitude control system

Nano-satellites are essential tools for various applications, including scientific experiments, deep space exploration and astronomical observation. Achieving precise model predictions is crucial for their successful operation. To address the intricate constraints of nano-satellites and enhance control performance, the Model Predictive Control (MPC) algorithm is an effective solution. However, implementing an MPC-based attitude control system in actual engineering scenarios presents significant challenges, primarily due to the substantial computational burden, especially given the limited onboard computing resources of nano-satellites. In this paper, we introduce a modified adaptive self-triggered model predictive control (ST-MPC) algorithm designed to stabilize the attitude of nano-satellites, while simultaneously reducing communication and computational overhead compared to traditional MPC methods. The proposed self-triggered mechanism dynamically determines the next trigger time according to the system state. Moreover, we incorporate considerations for the efficiency of actuators to address the constraints imposed by the magnetic torque characteristics within the modified self-triggered mechanism. Additionally, a strategy for adaptive prediction horizon is proposed to balance computation load and control accuracy. The results of our simulations demonstrate the effectiveness of the modified ST-MPC algorithm in comparison to both traditional MPC and standard ST-MPC approaches. This algorithm may have the potential to significantly impact attitude control applications for nano-satellites.

Hidden data recovery using reservoir computing: Adaptive network model and experimental brain signals.

The problem of hidden data recovery is crucial in various scientific and technological fields, particularly in neurophysiology, where experimental data can often be incomplete or corrupted. We investigate the application of reservoir computing (RC) to recover hidden data from both model Kuramoto network system and real neurophysiological signals (EEG). Using an adaptive network of Kuramoto phase oscillators, we generated and analyzed macroscopic signals to understand the efficiency of RC in hidden signal recovery compared to linear regression (LR). Our findings indicate that RC significantly outperforms LR, especially in scenarios with reduced signal information. Furthermore, when applied to real EEG data, RC achieved more accurate signal reconstruction than traditional spline interpolation methods. These results underscore RC's potential for enhancing data recovery in neurophysiological studies, offering a robust solution to improve data integrity and reliability, which is essential for accurate scientific analysis and interpretation.