Tuning Process Research Articles

Heart disease prediction utilizing machine learning techniques

Heart disease is an international public health issue and a significant health risk for many people. The World Health Organization (WHO) reported that heart disease has been identified as one of the primary causes of death. Owing to its rapid development and application in many areas, Machine Learning has become an effective technique for predicting heart disease. Machine learning, along with large-scale medical data and advanced algorithms, can assist healthcare professionals in accurately predicting the risk of heart disease, thus providing early intervention and treatment for patients. This research paper uses the “heart_2020_cleaned.csv” dataset, containing 319795 instances and 18 attributes, of which 70% of instances were randomly selected for the training set and 30% for testing. Applying machine learning algorithms in data mining such as Decision Tree (DT), LightGBM, Random Forest (RF) and Logistic Regression (LR) to forecast heart disease. Before constructing models, data cleaning, feature selection and hyperparameter tuning processes were done, aiming to explore the potential patterns among data. Comparative Analysis was conducted on the external test set to compare the prediction performance of different models at the same level. The result reported that the highest accuracy achieved with LightGBM was 76.9%, followed by Logistic Regression and Random Forest, with Decision Tree being the worst.

FI-NPI: Exploring Optimal Control in Parallel Platform Systems

Typically, the current and speed loop closure of servo motor of the parallel platform is accomplished with incremental PI regulation. The control method has strong robustness, but the parameter tuning process is cumbersome, and it is difficult to achieve the optimal control state. In order to further optimize the performance, this paper proposes a double-loop control structure based on fuzzy integral and neuron proportional integral (FI-NPI). The structure makes full use of the control advantages of the fuzzy controller and integrator to improve the performance of speed closed-loop control. And through the feedforward branch, the speed error is used as the teacher signal for neuron supervised learning, which improves the effect of current closed-loop control. Through comparative simulation experiments, this paper verifies that the FI-NPI controller has a faster dynamic response speed than the traditional PI controller. Finally, in this paper, the FI-NPI controller is implemented in C language in the servo-driven lower computer, and the speed closed-loop test of the BLDC motor is carried out. The experimental results show that the FI-NPI double-loop controller is better than the traditional double-PI controller in performance indicators such as convergence rate and RMSE, which confirms that the FI-NPI double-loop controller is more suitable for BLDC servo control.

Analysis Of One-Class and Binary Classification For COVID-19 Image Data

With the recent Coronavirus Disease 2019 (COVID-19) outbreak, there has been a growing interest in the utilization of machine learning models to aid in medical diagnostics. However, during the initial phases of the pandemic, the available datasets were severely imbalanced due to a scarcity of positive cases. These datasets predominantly comprised normal cases, with only a limited number of disease instances. This inherent data imbalance restricted the effectiveness of binary classification, prompting an exploration of one-class classification. This study aims to compare the performance of one-class and binary classification models during the early stages of COVID-19 development. The objective is to provide healthcare professionals with the most effective machine learning-assisted solutions for handling future outbreaks of a similar nature. The experiment incorporates three distinct models: One-Class Support Vector Machine (OCSVM) for one-class classification, alongside Categorical Boosting (CatBoost) and Extreme Gradient Boosting (XGBoost) for binary classification. These models undergo a rigorous tuning process employing grid search, using training data composed of features extracted from compressed chest X-ray images. In summary, the research findings unequivocally demonstrate the superior performance of one-class classification over binary classification in all early stages, achieving an impressive F1 score of 0.46, compared to binary classification's modest score of 0.063. Consequently, it is strongly recommended to employ a one-class classification model to enhance real-life diagnostic processes. However, it is anticipated that as more positive cases are incorporated into the dataset, achieving a more balanced distribution, binary classification may exhibit greater potential. Thus, medical practitioners should reconsider the applicability of binary classification in later stages when the dataset attains improved balance.

KONSEP NYUPAK DALAM KARAWITAN JAWA THE CONCEPT OF ‘NYUPAK’ IN JAVANESE TRADITIONAL MUSIC INSTRUMEN

The word ‘nyupak’ is a Javanese term which is often used in daily events or activities. ‘Nyupak’ is simply an activity of joining two objects so that the main object joined with another object can be used again according to its function. As a matter of fact, the term ‘nyupak’ is also used in karawitan organology in the process of (1) the tuning of ‘gender bumbungan’ (‘gender’ resonance) and (2) the tuning of gamelan, both in the form of blades and in the form of ‘pencon’. The current study aims at exploring the concept of ‘nyupak’ in Javanese traditional music instrument from acoustic perspective. The results of the study demonstrate that ‘nyupak’ in the tuning on ‘gender bumbungan’ (‘gender’ resonance) ranges from the intervals of 2 cents up to 12 cents, or written as 2 ≤ x ≤ 12 cents, and ‘nyupak’ in the tuning on gamelan ranges from the intervals of 10 cents up to 33 cents, or written as 10 ≤ x ≤33 cents. The result of the study provides a referential formulation for the tuning process of ‘gender bumbungan’ and gamelan, particularly to cope with ear sensitivity in capturing the tone frequency.

Research on a Vibrationally Tuned Directional Seed Supply Method Based on ADAMS-EDEM Coupling and the Optimization of System Parameters

We have combined the theory of bulk dynamics and the agronomic requirements of precision sowing with the aim of resolving the technical problems of poor seed mobility and the difficulty in controlling suction posture, which leads to an increase in the leakage rate and a reduction in seed qualification index scores. In this study, a vibrationally tuned directional seed supply method and system are proposed. We carried out a force analysis of seeds, constructed kinematic equations for seeds and seed boxes to specify the state of the seed motion, and determined the structural parameters and the range of structural parameters that affect the seed suction posture. In addition, we coupled the ADAMS-EDEM simulation of the motion process of the seed and seed boxes and analyzed the vibrational tuning process of the seeds and the angle of inclination of the bottom surface of the seed box. The speed of the eccentric wheel and the eccentric distance were used as test factors. Three-factor and three-level Box–Behnken central combination testing with a single-grain rate, multiple-grain rate, and cavity rate were used as response indicators. Mathematical models were obtained between the experimental factors and the response indicators. Multi-objective optimization of mathematical regression models was carried out with Design-Expert 10.0.4 software. The optimal parameter combination obtained was a tilt angle of 14.27°, an eccentric wheel speed of 4.48 rad/s, and an eccentricity of 1.94 mm. The rate of single grains was 90.75%, the rate of multiple grains was 3.63%, and the rate of cavities was 5.62%. In bench performance tests, using an angle of inclination of 14°, the speed of the eccentric wheel was 4.50 rad/s and the eccentricity was 2 mm. The mean value of the single-grain rate was 89.28%, the mean value of the multiple-grain rate was 3.89%, and the mean value of the cavity rate was 6.83%. The test error was within permissible limits, and reliable results were achieved for parameter optimization. The results met the technical requirements for precision sowing. The results of the study can provide academic references for theoretical research on the methodology of posturing and directional seed supply. They can also provide ideas for the design and development of seed supply systems for precision sowing machinery.

Prediction of chlorine residual in effluent based on SSA-SFA-CFBLS for water treatment process parameters

Accurate parameter prediction of chlorine residual in effluent is necessary for nonlinear, long-delayed water treatment process to raise water quality. To improve the prediction model precision and computational efficiency, a cascade broad learning system based on the sparrow search algorithm (SSA) and slow feature analysis (SFA) is proposed in this paper. First, the SFA method is introduced to extract the essential characteristics of water monitoring data as the input of the prediction model. Then, a cascaded broad learning system is adopted to establish a prediction model for residual chlorine in water works effluent. The cascade broad learning can deal well with online prediction. Furthermore, the SSA is utilized to obtain the optimal hyperparameters of the established model, which can avoid the complex and time-consuming manual parameter tuning process. Finally, the comparison experiment with several methods is carried out. The experimental results show that the proposed method saves more computational resources, and its time consumption is only 16.8% of that of the comprehensive deep learning model with basically the same accuracy. The prediction accuracy is improved, and its prediction accuracy is improved by 7.6% on average compared with most traditional deep learning models such as long short term memory (LSTM), recurrent neural network (RNN) and fated recurrent unit (GRU).

Tunable narrow linewidth fiber laser based on parity-time symmetric double rings and variable optical attenuator

A tunable narrow linewidth fiber ring laser based on parity-time (PT) symmetric double rings and variable optical attenuator (VOA) is demonstrated. The laser is composed of a fiber loop mirror (FLM), which creates two coupled rings: the reflective ring and the transmissive ring. The loss of the transmissive ring can be controlled by tuning the VOA, which balances with the gain of the reflective ring. When the PT symmetry is broken, a single-longitudinal-mode (SLM) laser is realized. The laser can be tuned by an optical tunable filter (OTF), and the tuning range is 38 nm, spanning from 1525 nm to 1563 nm. In the whole tuning process, the linewidth of the laser is at the sub-kHz level, and the optical signal to noise ratios (OSNR) is over 53 dB. The fluctuation of wavelength and power are less than 0.02 nm and 0.52 dB in an hour, respectively.

FAOT-Net: A 1.5-Stage Framework for 3D Pelvic Lymph Node Detection With Online Candidate Tuning.

Accurate and automatic detection of pelvic lymph nodes in computed tomography (CT) scans is critical for diagnosing lymph node metastasis in colorectal cancer, which in turn plays a crucial role in its staging, treatment planning, surgical guidance, and postoperative follow-up of colorectal cancer. However, achieving high detection sensitivity and specificity poses a challenge due to the small and variable sizes of these nodes, as well as the presence of numerous similar signals within the complex pelvic CT image. To tackle these issues, we propose a 3D feature-aware online-tuning network (FAOT-Net) that introduces a novel 1.5-stage structure to seamlessly integrate detection and refinement via our online candidate tuning process and takes advantage of multi-level information through the tailored feature flow. Furthermore, we redesign the anchor fitting and anchor matching strategies to further improve detection performance in a nearly hyperparameter-free manner. Our framework achieves the FROC score of 52.8 and the sensitivity of 91.7% with 16 false positives per scan on the PLNDataset. Code will be available at: github.com/SCUsomebody/FAOT-Net/.

A comparative study of modeled and measured room responses

At the heart of Brown University’s Lindemann Performing Arts Center is a shape-shifting hall capable of supporting a wide range of performance types. Acoustic simulation played a crucial role in optimizing each of the hall’s five primary configurations during design. Leading up to the opening events 2023 and 2024, the lengthy commissioning and tuning process provided an invaluable opportunity to measure subtle and drastic changes in the hall’s form and materiality for comparison to the predictive simulation. Drawing from the Lindemann set, the paper compares modeled and measured data from the PAC to that from other projects designed using similar tools. The paper identifies points of uncertainty during design and delves into commonalities and differences between ray-based simulation and measurement in performance spaces of varying form. This paper will address the inherent complexities of approximating acoustic behavior with side by side comparison of measurement and simulated data, we will share insights that inform our future modeling process.

Adaptive Bayesian contextual hyperband: A novel hyperparameter optimization approach

<p>Hyperparameter tuning plays a significant role when building a machine learning or a deep learning model. The tuning process aims to find the optimal hyperparameter setting for a model or algorithm from a pre-defined search space of the hyperparameters configurations. Several tuning algorithms have been proposed in recent years and there is scope for improvement in achieving a better exploration-exploitation tradeoff of the search space. In this paper, we present a novel hyperparameter tuning algorithm named adaptive Bayesian contextual hyperband (Adaptive BCHB) that incorporates a new sampling approach to identify best regions of the search space and exploit those configurations that produce minimum validation loss by dynamically updating the threshold in every iteration. The proposed algorithm is assessed using benchmark models and datasets on traditional machine learning tasks. The proposed Adaptive BCHB algorithm shows a significant improvement in terms of accuracy and computational time for different types of hyperparameters when compared with state-of-the-art tuning algorithms.</p>

TSFIS-GWO: Metaheuristic-driven takagi-sugeno fuzzy system for adaptive real-time routing in WBANs

Wireless body area network (WBAN) is an internet-of-things technology that facilitates remote patient monitoring and enables medical staff to administer timely treatments. One of the main challenges in designing WBANs is the routing problem, which is complicated due to dynamic changes in network topology and the limited resources of nodes. Several heuristic and metaheuristic methods have been presented to solve the routing problem in WBANs. Although metaheuristics outperform heuristics by producing higher-quality solutions, they cannot respond to real-time requests. This paper introduces a reactive routing protocol for WBANs that combines a fuzzy heuristic with a metaheuristic learning model. It utilizes a Takagi-Sugeno Fuzzy Inference System in conjunction with the Grey Wolf Optimizer (named TSFIS-GWO). The objective is to simultaneously benefit from the advantages of both approaches, namely, the effectiveness of metaheuristics for offline hyperparameter tuning and the quickness of fuzzy heuristics for real-time routing. At every round, the tuned fuzzy system takes multiple parameters of the current state of the nodes and links to construct the multi-hop routing tree under IEEE 802.15.6. To optimize the performance of the protocol for each WBAN, the fuzzy rules of the TSFIS model are automatically adjusted through a learning method based on GWO. This is done in accordance with the specific requirements of the application, and the tuning process takes place once before the protocol is applied. Simulation results in three applications demonstrate that the proposed TSFIS-GWO model is capable of providing real-time solutions while outperforming the existing methods in terms of application-specific performance measures.

Archimedes optimisation algorithm quantum dilated convolutional neural network for road extraction in remote sensing images

Roads are closely intertwined with human existence, and the process of extracting road networks has emerged as the most prominent task in remote sensing (RS). The automated road interpretation process of remote sensing images (RSI) efficiently acquires road network data at a reduced expense in comparison to the traditional visual interpretation of RSI. However the manifestation of RSI is completely distinct because of the great difference in length, width, material, and shape of road networks in dissimilar areas. Thus, the extraction of road network data in RSI is still a complex issue. In recent times, DL-based approaches have projected a famous development in image segmentation outcomes, but a lot of them still could not retain boundary data and attain high-resolution road segmentation maps while processing the RSI. Traditional convolutional neural networks (CNNs) demonstrate impressive performance in road extract tasks; however, they frequently encounter difficulties in capturing intricate details and contextual information. The study introduces a novel method, named Archimedes Optimisation Algorithm, Quantum Dilated Convolutional Neural Network for Road Extraction (AOA-QDCNNRE), to tackle the challenges encountered in remote sensing images. The AOA-QDCNNRE technique aims to generate a high-resolution road segmentation map using DL with a hyperparameter tuning process. The AOA-QDCNNRE technique primarily relies on the QDCNN model, which integrates quantum technology (QC) with dilated convolutions to augment the network's capacity to capture local as well as global contextual information. In addition, the incorporation of the dilated convolutional technique effectively enhances the receptive field without sacrificing spatial resolution, enabling the extraction of precise road features. To develop the road extraction outcomes of the QDCNN approach, the AOA-based hyperparameter tuning process can be exploited. The AOA-QDCNNRE system's simulation results can be tested on benchmark databases, and the results indicate that the AOA-QDCNNRE method surpasses recent algorithms.

A fuzzy computing approach to aggregate expert opinions using parabolic and exparabolic approximation procedures for solving multi-criteria group decision-making problems

AbstractTriangular fuzzy numbers (TFNs) are widely used for selection problems to determine expert opinions using linguistic expressions. Some aggregation procedures are developed to determine expert opinions more accurately. However, there is a need for a simple and more useful procedure to solve the selection problems more suitably. For this purpose, our study offers a triangular, exparabolic, and parabolic area calculation-based approximation approach for TFNs to aggregate the possible hedges (very and more or less) for TFNs. Hence, this aggregation procedure provides a tuning opportunity for classical TFN expressions to capture possible tuning processes to reflect the hesitancies of experts. The technique for order preferences by similarity to ideal solution (TOPSIS) method is applied in the two studies from extant literature, and suitable alternatives are determined as a result of the ranking process. Finally, a comparative analysis is presented to illustrate the efficiency of the proposed procedure. The conventional TOPSIS model’s ranking scores are very close for exemplified examples (i.e., 0.5308, 0.4510, 0.4550 and 0.5304, 0.4626, 0.4940), but the proposed model’s result has fluctuated for the same examples (i.e., 0.346, 0,669, 0,567 and 0.208, 0.991, 0.148). So, the main advantage of the proposed aggregation procedure is the alternative ranking scores separation capability analyzed with their linguistic diversification.

Disturbance rejection of offshore drilling platforms: An equivalent-input-disturbance-based dynamic positioning control method

Offshore drilling platforms are exposed to wind, waves, currents, and other unknown disturbances. Accurately estimating and rejecting these disturbances is the key to ensuring reliable station-keeping of the platforms. In this study, a novel dynamic positioning method using an improved equivalent-input-disturbance (EID) approach is proposed for offshore drilling platforms. An improved EID estimator is employed to estimate and suppress unknown disturbances, significantly enhancing the disturbance-rejection performance of the dynamic positioning system. The input channels are decoupled through linear transformation, and the parameter tuning process of the observer and controller is optimized, thus improving system performance. The bounded-input bounded-output stability of the closed-loop system is proved. This study provides insights into the design of dynamic positioning systems for offshore drilling platforms.

Optimal design of I-PD controller for disturbance rejection of time delayed unstable and integrating-unstable processes

Disturbance rejection has always been a major phenomenon in control theory. Disturbances that arise in control of unstable or integrating unstable processes with time delay present considerable difficulties for classical PID controllers. This paper supplies analytical tuning rules, derived from optimum disturbance rejection responses to minimise the error signal according to several integral performance criteria to identify the tuning parameters of the I-PD controller. The provided analytical rules offer the advantage of calculating controller parameters without the necessity of employing an optimisation algorithm. This simplifies the tuning process and allows for a straightforward determination of the controller's parameters, making it more convenient and efficient for practical implementation in control systems. Comprehensive simulations were performed to validate the effectiveness of the proposed I-PD controller in terms of disturbance rejection responses, control signals, perturbation in process parameters, measurement noises, TV values, Ms values, and integral performance indicators. Overall, the outcomes demonstrate that the introduced method for the tuning of I-PD controllers offers notable advantages when compared to other tuning methods found in the literature.

Gait quality in prosthesis users is reflected by force-based metrics when learning to walk on a new research-grade powered prosthesis.

Powered prosthetic feet require customized tuning to ensure comfort and long-term success for the user, but tuning in both clinical and research settings is subjective, time intensive, and the standard for tuning can vary depending on the patient's and the prosthetist's experience levels. Therefore, we studied eight different metrics of gait quality associated with use of a research-grade powered prosthetic foot in seven individuals with transtibial amputation during treadmill walking. We compared clinically tuned and untuned conditions with the goal of identifying performance-based metrics capable of distinguishing between good (as determined by a clinician) from poor gait quality. Differences between the tuned and untuned conditions were reflected in ankle power, both the vertical and anterior-posterior impulse symmetry indices, limb-force alignment, and positive ankle work, with improvements seen in all metrics during use of the tuned prosthesis. Notably, all of these metrics relate to the timing of force generation during walking which is information not directly accessible to a prosthetist during a typical tuning process. This work indicates that relevant, real-time biomechanical data provided to the prosthetist through the future provision of wearable sensors may enhance and improve future clinical tuning procedures associated with powered prostheses as well as their long-term outcomes.

Optimization of Patient Specific Stimulus for Deep Brain Stimulation Using Spatially Distributed Neural Sources

Deep brain stimulation (DBS) becomes the therapy of choice in the later stages of Parkinson's disease (PD) due to the medication's side effects. For effective DBS treatment, it is important to have a controlled dosage of DBS. DBS dosage is administered using the tuning of electrical parameters of the stimulus signal. Since this tuning process is tedious, time-consuming and patient-specific, there is a need to study the properties of DBS stimulation signal for proper dose administration. We propose a simulation framework to define an optimized DBS stimulus using Electroencephalogram (EEG) signals. The objective is to provide a simulation environment inspired by a realistic brain. The framework uses spiking neurons in a reservoir modelled after real brain anatomy and is trained using a biologically inspired spike-time-dependent-plasticity learning algorithm. This reservoir is initially set to OFF-medication state and forced to drift to the ON-medication state by optimizing the synaptic changes. In later testing, the generalization of this framework is verified with EEG-inverse solutions like sLORETA which utilize time-domain EEG signals to estimate neural activations. The stimulus signal is generated by accumulating the variations in synaptic weights in the neural reservoir in the target brain region. We analyze this signal and show that the application of this signal as stimulus results in decreased <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\beta$</tex-math></inline-formula> -band power in subthalamic-nucleus local-field-potential compared to OFF-medication local-field-potential without stimulation. Using SIM4LIFE simulation software, we show that the simulation increases chaos in the local-field-potential of subthalamic-nucleus neurons and shows that neuron weight variations follow specific trajectories in reconstructed state-space. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Impact Statement:</i> The primary contribution of this paper is to present a framework that utilizes the association between target signals of any modality (EEG, MEG, fMRI) and their respective post-synaptic activations to optimize the deep brain stimulation (DBS) stimulus signal. This framework combines coordinates of neural sources with patients' EEG data to optimize the stimulus of DBS. While defining this framework, we initially proposed the usage of spiking neurons in a biological neural reservoir to estimate the internal activations for given EEG data. Later, we extended our testing to incorporate inverse EEG algorithms that estimate neural activations directly from EEG data without any conversion. This is a novel framework, and there are not many examples available in the literature for similar work. This framework is in its early stages, yet it provides an adequate starting point for an under-explored area of study. This framework can potentially evolve into a useful tool for doctors in assisting the DBS tuning process before the surgical procedure. This framework's accuracy depends on the underlying technique to estimate the neural activations from the target signals (like EEG etc.).

A Study on Fractional Power-Law Applications and Approximations

The frequency response of the fractional-order power-law filter can be approximated by different techniques, which eventually affect the expected performance. Fractional-order control systems introduce many benefits for applications like compensators to achieve robust frequency and additional degrees of freedom in the tuning process. This paper is a comparative study of five of these approximation techniques. The comparison focuses on their magnitude error, phase error, and implementation complexity. The techniques under study are the Carlson, continued fraction expansion (CFE), Padé, Charef, and MATLAB curve-fitting tool approximations. Based on this comparison, the recommended approximation techniques are the curve-fitting MATLAB tool and the continued fraction expansion (CFE). As an application, a low-pass power-law filter is realized on a field-programmable analog array (FPAA) using two techniques, namely the curve-fitting tool and the CFE. The experiment aligns with and validates the numerical results.

Beam control system and output fine-tuning for safe and precise delivery of FLASH radiotherapy at a clinical linear accelerator.

We have previously adapted a clinical linear accelerator (Elekta Precise, Elekta AB) for ultra-high dose rate (UHDR) electron delivery. To enhance reliability in future clinical FLASH radiotherapy trials, the aim of this study was to introduce and evaluate an upgraded beam control system and beam tuning process for safe and precise UHDR delivery. The beam control system is designed to interrupt the beam based on 1) a preset number of monitor units (MUs) measured by a monitor detector, 2) a preset number of pulses measured by a pulse-counting diode, or 3) a preset delivery time. For UHDR delivery, an optocoupler facilitates external control of the accelerator's thyratron trigger pulses. A beam tuning process was established to maximize the output. We assessed the stability of the delivery, and the independent interruption capabilities of the three systems (monitor detector, pulse counter, and timer). Additionally, we explored a novel approach to enhance dosimetric precision in the delivery by synchronizing the trigger pulse with the charging cycle of the pulse forming network (PFN). Improved beam tuning of gun current and magnetron frequency resulted in average dose rates at the dose maximum at isocenter distance of >160 Gy/s or >200 Gy/s, with or without an external monitor chamber in the beam path, respectively. The delivery showed a good repeatability (standard deviation (SD) in total film dose of 2.2%) and reproducibility (SD in film dose of 2.6%). The estimated variation in DPP resulted in an SD of 1.7%. The output in the initial pulse depended on the PFN delay time. Over the course of 50 measurements employing PFN synchronization, the absolute percentage error between the delivered number of MUs calculated by the monitor detector and the preset MUs was 0.8 ± 0.6% (mean ± SD). We present an upgraded beam control system and beam tuning process for safe and stable UHDR electron delivery of hundreds of Gy/s at isocenter distance at a clinical linac. The system can interrupt the beam based on monitor units and utilize PFN synchronization for improved dosimetric precision in the dose delivery, representing an important advancement toward reliable clinical FLASH trials.

Deep learning model for dynamic color design of all-dielectric metasurfaces.

Silicon nanostructure colors have rapidly developed in recent years, offering high resolution and a broad color gamut that traditional pigments cannot achieve. The reflected colors of metasurfaces are determined by the geometric structure of the unit cell and the refractive index matching layer parameters. It is evident that the design of specific colors involves numerous parameters, making it challenging to achieve through conventional calculations. Therefore, the tandem network instead of conventional electromagnetic simulation is natural. The forward part of the network incorporates feature cross terms to improve accuracy, enabling high-precision predictions of structural colors based on structural parameters. The average color difference between the predicted and actual color values in the L,a,b color space is 1.38. The network has been proven to accurately predict the refractive index and height of the refractive index matching layer during the dynamic tuning process. Additionally, the issue of the inverse network converging to incorrect solutions was addressed by leveraging the characteristics of the activation function. The results show that the color difference between the colors designed by the inverse network compared to the actual colors in the L,a,b color spaces is only 2.22, which meets the requirements for commercial applications.