Parameter Tuning Techniques Research Articles

An explainable machine learning approach for automated medical decision support of heart disease

Coronary Heart Disease (CHD) is the dominant cause of mortality around the world. Every year, it causes about 3.9 million deaths in Europe and 1.8 million in the European Union (EU). It is responsible for 45 % and 37 % of all deaths in Europe and the European Union, respectively. Using machine learning (ML) to predict heart diseases is one of the most promising research topics, as it can improve healthcare and consequently increase the longevity of people's lives. However, although the ability to interpret the results of the predictive model is essential, most of the related studies do not propose explainable methods. To address this problem, this paper presents a classification method that not only exhibits reliable performance but is also interpretable, ensuring transparency in its decision-making process. SHapley Additive exPlanations, known as the SHAP method was chosen for model interpretability. This approach presents a comparison between different classifiers and parameter tuning techniques, providing all the details necessary to replicate the experiment and help future researchers working in the field. The proposed model achieves similar performance to those proposed in the literature, and its predictions are fully interpretable.

Medical Support Vehicle Location and Deployment at Mass Casualty Incidents

Anticipating and planning for the urgent response to large-scale disasters is critical to increase the probability of survival at these events. These incidents present various challenges that complicate the response, such as unfavorable weather conditions, difficulties in accessing affected areas, and the geographical spread of the victims. Furthermore, local socioeconomic factors, such as inadequate prevention education, limited disaster resources, and insufficient coordination between public and private emergency services, can complicate these situations. In large-scale emergencies, multiple demand points (DPs) are generally observed, which requires efforts to coordinate the strategic allocation of human and material resources in different geographical areas. Therefore, the precise management of these resources based on the specific needs of each area becomes fundamental. To address these complexities, this paper proposes a methodology that models these scenarios as a multi-objective optimization problem, focusing on the location-allocation problem of resources in Mass Casualty Incidents (MCIs). The proposed case study is Mexico City in a earthquake post-disaster scenario, using voluntary geographic information, open government data, and historical data from the 19 September 2017 earthquake. It is assumed that the resources that require optimal location and allocation are ambulances, which focus on medical issues that affect the survival of victims. The designed solution involves the use of a metaheuristic optimization technique, along with a parameter tuning technique, to find configurations that perform at different instances of the problem, i.e., different hypothetical scenarios that can be used as a reference for future possible situations. Finally, the objective is to present the different solutions graphically, accompanied by relevant information to facilitate the decision-making process of the authorities responsible for the practical implementation of these solutions.

Potentiality of automatic parameter tuning suite available in ACTS track reconstruction software framework

Particle tracking is among the most sophisticated and complex part of the full event reconstruction chain. A number of reconstruction algorithms work in a sequence to build these trajectories from detector hits. Each of these algorithms use many configuration parameters that need to be fine-tuned to properly account for the detector/experimental setup, the available CPU budget and the desired physics performance. Few examples of such parameters include the cut values limiting the search space of the algorithm, the approximations accounting for complex phenomena or the parameters controlling algorithm performance. The most popular method to tune these parameters is hand-tuning using brute-force techniques. These techniques can be inefficient and raise issues for the long-term maintainability of such algorithms. The opensource track reconstruction software framework known as “A Common Tracking Framework (ACTS)” offers an alternative solution to these parameter tuning techniques through the use of automatic parameter optimization algorithms. ACTS come equipped with an auto-tuning suite that provides necessary setup for performing optimization of input parameters belonging to track reconstruction algorithms. The user can choose the tunable parameters in a flexible way and define a cost/benefit function for optimizing the full reconstruction chain. The fast execution speed of ACTS allows the user to run several iterations of optimization within a reasonable time bracket. The performance of these optimizers has been demonstrated on different track reconstruction algorithms such as trajectory seed reconstruction and selection, particle vertex reconstruction and generation of simplified material map, and on different detector geometries such as Generic Detector and Open Data Detector (ODD). We aim to bring this approach to all aspects of trajectory reconstruction by having a more flexible integration of tunable parameters within ACTS.

An efficient network intrusion detection approach based on logistic regression model and parallel artificial bee colony algorithm

In recent years, the widespread use of the Internet has created many issues, especially in the area of cybersecurity. It is critical to detect intrusions in network traffic, and researchers have developed network intrusion and anomaly detection systems to cope with high numbers of attacks and attack variations. In particular, machine learning and meta-heuristic methods have been widely used for network intrusion detection systems (NIDS). However, existing studies on these systems usually suffer from low performance results such as accuracy, F1-measure, false positive rate, and false negative rate, and generally do not use automatic parameter tuning techniques. To address these challenges, this study proposes a novel approach based on a logistic regression model trained using a parallel artificial bee colony (LR-ABC) algorithm with a hyper-parameter optimization technique. The performance of the proposed model is evaluated against state-of-the-art machine learning and deep learning models on two publicly available NIDS datasets. Comparative performance evaluations show that the proposed method achieved satisfactory results with accuracy of 88.25% on the UNSW-NB15 dataset and 90.11% on the NSL-KDD dataset, and F1-measures of 88.26% and 90.15%, respectively. These findings demonstrate the efficacy of the proposed LR-ABC model in enhancing the accuracy and reliability, while providing a scalable solution to adapt to the dynamic and evolving landscape of cybersecurity threats.

Training Artificial Neural Networks Using a Global Optimization Method That Utilizes Neural Networks

Perhaps one of the best-known machine learning models is the artificial neural network, where a number of parameters must be adjusted to learn a wide range of practical problems from areas such as physics, chemistry, medicine, etc. Such problems can be reduced to pattern recognition problems and then modeled from artificial neural networks, whether these problems are classification problems or regression problems. To achieve the goal of neural networks, they must be trained by appropriately adjusting their parameters using some global optimization methods. In this work, the application of a recent global minimization technique is suggested for the adjustment of neural network parameters. In this technique, an approximation of the objective function to be minimized is created using artificial neural networks and then sampling is performed from the approximation function and not the original one. Therefore, in the present work, learning of the parameters of artificial neural networks is performed using other neural networks. The new training method was tested on a series of well-known problems, a comparative study was conducted against other neural network parameter tuning techniques, and the results were more than promising. From what was seen after performing the experiments and comparing the proposed technique with others that have been used for classification datasets as well as regression datasets, there was a significant difference in the performance of the proposed technique, starting with 30% for classification datasets and reaching 50% for regression problems. However, the proposed technique, because it presupposes the use of global optimization techniques involving artificial neural networks, may require significantly higher execution time than other techniques.

Efficient parameter tuning to enhance practicability of a model-free vibration controller based on a virtual controlled object

In active vibration control, controller tuning is necessary to obtain a sufficient damping performance. This study presents an efficient tuning technique for model-free vibration controller based on a virtual controlled object (VCO). A model-free vibration control system is constructed by using an actuator model and the VCO. A reference controlled object (RCO), which is a designer-defined single-degree-of-freedom (SDOF) structure, is used to tune the VCO-controller offline. A novel parameter tuning technique based on the RCO and the simultaneous perturbation stochastic approximation (SPSA) is proposed considering the difference in scale of the tuning parameters. The proposed tuning scheme automatically determines the tuning parameter of the controller without manual trial-and-errors such as experiments with actual plants and is much faster than previous methods. Therefore, the proposed method significantly enhances the practicability of the VCO-based model-free vibration suppression, bridging the gap between the basic study of the VCO-method and its implementation. The effectiveness of the proposed method is verified by applying it to the VCO-based H2 controller.

Lung Cancer Classification in Histopathology Images Using Multiresolution Efficient Nets.

Histopathological images are very effective for investigating the status of various biological structures and diagnosing diseases like cancer. In addition, digital histopathology increases diagnosis precision and provides better image quality and more detail for the pathologist with multiple viewing options and team annotations. As a result of the benefits above, faster treatment is available, increasing therapy success rates and patient recovery and survival chances. However, the present manual examination of these images is tedious and time-consuming for pathologists. Therefore, reliable automated techniques are needed to effectively classify normal and malignant cancer images. This paper applied a deep learning approach, namely, EfficientNet and its variants from B0 to B7. We used different image resolutions for each model, from 224 × 224 pixels to 600 × 600 pixels. We also applied transfer learning and parameter tuning techniques to improve the results and overcome the overfitting problem. We collected the dataset from the Lung and Colon Cancer Histopathological Image LC25000 image dataset. The dataset acquisition consists of 25,000 histopathology images of five classes (lung adenocarcinoma, lung squamous cell carcinoma, benign lung tissue, colon adenocarcinoma, and colon benign tissue). Then, we performed preprocessing on the dataset to remove the noisy images and bring them into a standard format. The model's performance was evaluated in terms of classification accuracy and loss. We have achieved good accuracy results for all variants; however, the results of EfficientNetB2 stand excellent, with an accuracy of 97% for 260 × 260 pixels resolution images.

Simulation-based analysis of second-best multimodal network capacity

Modeling the capacity of a transportation system ought to be an essential task for assessing the level of service of urban transportation networks. Compared to the maximum traffic throughput under idealized conditions, the total network origin–destination (OD) demand considering practical and political requirements has recently been brought into focus. This problem can be referred to as the second-best network capacity (SNC) analysis and is fully investigated in this paper. To quantify the SNC of multimodal networks, an accurate traffic assignment model incorporating combined modes such as Park-and-Ride is necessary. This paper develops a simulation-based approach that exploits the advantages of the off-the-shelf simulator to reduce model complexity. A bilevel formulation of SNC analysis in a simulation-based manner is proposed to obtain optimal OD demand patterns. Owing to the black-box nature of simulation models, there is a need for computationally efficient algorithms dispensing with gradient approximations. Inspired by machine learning parameter tuning techniques, a Bayesian-type algorithm is employed to solve the model, which can considerably save the number of simulation evaluations. Case studies are conducted on two networks: the Braess and the Sioux Falls networks. The results suggest that the proposed method can accurately and robustly solve the complex SNC problem.

A New Positioning Strategy Based on Parameter Tuning and Optimal Control Technique for Pneumatic Control Valve

To improve the accuracy, rapidity, and versatility of pneumatic control valves, mechanism modeling, and optimization analysis on pneumatic actuators were carried out, and then a new positioning strategy based on parameter tuning and optimal control technique was proposed consequently. Firstly, a dynamic model of valve position based on optimal control under constraint conditions was established, and the characteristic parameters of valve position motion were determined. Then, the adaptive processing of switching points was introduced into the optimal control strategy to improve the versatility of the control. Finally, according to different control strategy partitions, the interactive control of the piezoelectric air switch by fully open control and PWM improves the control accuracy of the algorithm. The experimental analysis of the control effect with other brand positioners shows that the proposed control strategy shortens the system adjustment time, improves the accuracy of valve position, and avoids the overshoot.

Predicting Academic Performance of Immigrant Students Using XGBoost Regressor

The education sector has been effectively dealing with the prediction of academic performance of the Immigrant students since the research associated with this domain proves beneficial enough for those countries where the ministry of education has to cater to such immigrants for altering and updating policies in order to elevate the overall education pedagogy for them. The present research begins with analyzing varied educational data mining and machine learning techniques that helps in assessing the data fetched form PISA. It’s elucidated that XGBoost stands out to be the ideal most machine learning technique for achieving the desired results. Subsequently, the parameters have been optimized using the hyper parameter tuning techniques and implemented on the XGBoost Regressor algorithm. Resultant there is low error rate and higher level of predictive ability using the machine learning algorithms which assures better predictions using the PISA data. The final results have been discussed along with the upcoming future research work.

Robust longitudinal motion control of underground mining electric vehicles based on fuzzy parameter tuning sliding mode controller

The desire to reduce costs and increase productivity is driving mining industries to deploy underground mining electric vehicles (UMEVs). Due to the complex, working environments in underground mines, safer and more stable control strategies of UMEVs are required than the conventional control approaches used for electric vehicles (EVs). In this paper, first a UMEV model with bounded system uncertainties is derived. Next, a fuzzy parameter tuning (FPT) technique is used to adjust parameters in the sliding mode controller (SMC) against unexpected external disturbances and unbounded system uncertainties, which significantly improves system stability and robustness. Simulation and experimental results demonstrate that the proposed fuzzy parameter tuning sliding mode controller (FPTSMC) achieves better performance than SMCs with parametric uncertainties.

Metaheuristics with Deep Learning Empowered Biomedical Atherosclerosis Disease Diagnosis and Classification

Atherosclerosis diagnosis is an inarticulate and complicated cognitive process. Researches on medical diagnosis necessitate maximum accuracy and performance to make optimal clinical decisions. Since the medical diagnostic outcomes need to be prompt and accurate, the recently developed artificial intelligence (AI) and deep learning (DL) models have received considerable attention among research communities. This study develops a novel Metaheuristics with Deep Learning Empowered Biomedical Atherosclerosis Disease Diagnosis and Classification (MDL-BADDC) model. The proposed MDL-BADDC technique encompasses several stages of operations such as pre-processing, feature selection, classification, and parameter tuning. Besides, the proposed MDL-BADDC technique designs a novel Quasi-Oppositional Barnacles Mating Optimizer (QOBMO) based feature selection technique. Moreover, the deep stacked autoencoder (DSAE) based classification model is designed for the detection and classification of atherosclerosis disease. Furthermore, the krill herd algorithm (KHA) based parameter tuning technique is applied to properly adjust the parameter values. In order to showcase the enhanced classification performance of the MDL-BADDC technique, a wide range of simulations take place on three benchmarks biomedical datasets. The comparative result analysis reported the better performance of the MDL-BADDC technique over the compared methods.

IoT enabled environmental toxicology for air pollution monitoring using AI techniques

In past decades, the industrial and technological developments have increased exponentially and accompanied by non-judicial and un-sustainable utilization of non-renewable resources. At the same time, the environmental branch of toxicology has gained significant attention in understanding the effect of toxic chemicals on human health. Environmental toxic agents cause several diseases, particularly high risk among children, pregnant women, geriatrics and clinical patients. Since air pollution affects human health and results in increased morbidity and mortality increased the toxicological studies focusing on industrial air pollution absorbed by the common people. Therefore, it is needed to design an automated Environmental Toxicology based Air Pollution Monitoring System. To resolve the limitations of traditional monitoring system and to reduce the overall cost, this paper designs an IoT enabled Environmental Toxicology for Air Pollution Monitoring using Artificial Intelligence technique (ETAPM-AIT) to improve human health. The proposed ETAPM-AIT model includes a set of IoT based sensor array to sense eight pollutants namely NH3, CO, NO2, CH4, CO2, PM2.5, temperature and humidity. The sensor array measures the pollutant level and transmits it to the cloud server via gateways for analytic process. The proposed model aims to report the status of air quality in real time by using cloud server and sends an alarm in the presence of hazardous pollutants level in the air. For the classification of air pollutants and determining air quality, Artificial Algae Algorithm (AAA) based Elman Neural Network (ENN) model is used as a classifier, which predicts the air quality in the forthcoming time stamps. The AAA is applied as a parameter tuning technique to optimally determine the parameter values of the ENN model. In-order to examine the air quality monitoring performance of the proposed ETAPM-AIT model, an extensive set of simulation analysis is performed and the results are inspected in 5, 15, 30 and 60 min of duration respectively. The experimental outcome highlights the optimal performance of the proposed ETAPM-AIT model over the recent techniques.

Parameter tuning technique for a model-free vibration control system based on a virtual controlled object

A parameter tuning technique without manual trial-and-error procedures is proposed for a controller in a model-free vibration control system based on a virtual controlled object (VCO), which is defined as a single-degree-of-freedom (SDOF) system. The model-free control system is constructed by inserting a VCO between the actuator and the actual controlled object. A reference controlled object (RCO), which is also expressed as an SDOF system, is defined for the configured model-free control system. Then the loss function, which is calculated using the RCO vibration control simulation results, is used to evaluate the vibration suppression performance. The simultaneous perturbation stochastic approximation (SPSA) adjusts the controller tuning parameters to minimize the loss function. The SPSA- and RCO-based tuning procedures automatically tune the model-free controller without manual trial-and-error procedures. Simulations and experiments demonstrate that a model-free linear quadratic regulator designed by the proposed approach provides sufficient vibration reduction.

Performance analysis of PID control in DC Brushless motor using trial and error method

This research presents an overview of the performance of the PID control technique applied to the Brushless DC motor to regulate the speed using the Arduino microcontroller and LabView software. Brushless DC motors are chosen because they have high efficiency both in performance and longer durability. Kp, Ti and Td parameter tuning techniques use the Trial and error method and the results obtained from the value of Kp = 0.050, Ti = 0.001 and Td = 0.001 indicate a minimum error compared to controls without using PID. The percentage of overshoot still exists but the rise time and settling time indicate a faster response speed to the setpoint.

Adagrad Optimizer with Elephant Herding Optimization based Hyper Parameter Tuned Bidirectional LSTM for Customer Churn Prediction in IoT Enabled Cloud Environment

At recent times, customer churn is an important activity in quickly developing industries such as telecom, banking, e-commerce, etc. Earlier studies revealed that the cost of getting a new customer is considerably higher than the cost of retaining the existing ones. Therefore, it becomes essential to predict the nature of customer churn for retaining the customers to a greater extent. The advent of deep learning (DL) models have begun to be employed for efficient CCP. This paper presents a new Adagrad Optimizer with Elephant Herding Optimization (EHO) based Hyper-parameter Tuned Bidirectional Long Short Term Memory (AG-EHO-BiLSTM) for CCP in Internet of Things (IoT) enabled Cloud Environment. The proposed AG-EHO-BiLSTM model initially acquires the customer data using its devices like smart phones, laptop, smart watch, etc. Next, the gathered data will be classified by the use of Bi-LSTM model, which determines the customers as churner or non-churner. The efficiency of the Bi-LSTM model can be increased through hyper parameter tuning techniques, namely Adagrad optimizer and EHO algorithm to optimally select the parameter values namely learning rate, number of hidden layer and epochs. The performance validation of the AG-EHO-BiLSTM model takes place on benchmark dataset and the simulation outcome reported the supremacy of the AG-EHO-BiLSTM model over the comparative methods.

텍스트 마이닝과 딥 러닝을 활용한 고장진단: 고장원인부서 예측 알고리즘 개발

Recently, as the value of text data increases, various studies on text analysis have been required, especially in the area of failure diagnosis. However, there are almost no results that systematically address the text analysis problem in the literature. Motivated by this concern, this paper focuses on developing a text analysis method that can be applied to failure diagnosis via the well-known text mining and deep learning. The main objective of this study is to analyze the failure diagnosis data of electric equipments by building an algorithm that can predict the departments (or companies) that provides the cause of the failure. To be specific, the proposed algorithm offers a possibility 1) to shorten the maintenance time, 2) to minimize labor costs, and 3) provide clear criteria of selecting subcontractor through statics and visualization. Methodologically, the data preprocessing such as feature extraction and tokenization enables the raw text data to be analyzed. After that, on the basis of machine learning and deep learning models (including CNN and RNN), various experiments are conducted to select some suitable models with high performances. Furthermore, this study makes an attempt to optimize the selected models by means of parameter tuning and dropout techniques. Finally, using the most recent data, the effectiveness of the proposed algorithm is verified.

Assessing the impact of parameters tuning in ensemble based breast Cancer classification

Breast cancer is one of the major causes of death among women. Different decision support systems were proposed to assist oncologists to accurately diagnose their patients. These decision support systems mainly used classification techniques to categorize the diagnosis into Malign or Benign tumors. Given that no consensus has been reached on the classifier that can perform best in all circumstances, ensemble-based classification, which classifies patients by combining more than one single classification technique, has recently been investigated. In this paper, heterogeneous ensembles based on three well-known machine learning techniques (support vector machines, multilayer perceptron, and decision trees) were developed and evaluated by investigating the impact of parameter values of the ensemble members on classification performance. In particular, we investigate three parameters tuning techniques: Grid Search (GS), Particle Swarm Optimization (PSO) and the default parameters of the Weka Tool to evaluate whether setting ensemble parameters permits more accurate classification in breast cancer over four datasets obtained from the Machine Learning repository. The heterogeneous ensembles of this study were built using the majority voting technique as a combination rule. The overall results obtained suggest that: (1) Using GS or PSO techniques for single techniques provide more accurate classification; (2) In general, ensembles generate more accurate classification than their single techniques regardless of the optimization techniques used. (3) Heterogeneous ensembles based on optimized single classifiers generate better results than the Uniform Configuration of Weka (UC-WEKA) ensembles, and (4) PSO and GS slightly have the same impact on the performances of ensembles.

Machine-learning-assisted fabrication: Bayesian optimization of laser-induced graphene patterning using in-situ Raman analysis

The control of the physical, chemical, and electronic properties of laser-induced graphene (LIG) is crucial in the fabrication of flexible electronic devices. However, the optimization of LIG production is time-consuming and costly. Here, we demonstrate state-of-the-art automated parameter tuning techniques using Bayesian optimization to advance rapid single-step laser patterning and structuring capabilities with a view to fabricate graphene-based electronic devices. In particular, a large search space of parameters for LIG explored efficiently. As a result, high-quality LIG patterns exhibiting high Raman G/D ratios at least a factor of four larger than those found in the literature were achieved within 50 optimization iterations in which the laser power, irradiation time, pressure and type of gas were optimized. Human-interpretable conclusions may be derived from our machine learning model to aid our understanding of the underlying mechanism for substrate-dependent LIG growth, e.g. high-quality graphene patterns are obtained at low and high gas pressures for quartz and polyimide, respectively. Our Bayesian optimization search method allows for an efficient experimental design that is independent of the experience and skills of individual researchers, while reducing experimental time and cost and accelerating materials research.

Fine Tuned Deep Neural Networks for Intrusion Detection System

Intrusion detection system is used to gather and analyze data from networks for possible threats identification. It is used to record all significant events within network and takes corrective action when threats are detected. Some existing intrusion detection systems have experienced false positive or negative alarms. Recently, deep neural networks are being widely used to implement intrusion detection system to increase accuracy and reduce false alarms. Deep Neural networks are encouraged by the architecture depth of the human brain. These can possibly remove better portrayals from the information to make much better models. Tuning of hyper parameter of complex and computationally expensive deep networks is essential to advance classification accuracy by reducing training time. This hyper parameter tuning is also known as automated machine learning. In this paper, a deep neural network has been used for intrusion detection system. Different architecture of Multilayer perceptron has verified to learn best performance on intrusion detection system. Different hyper parameters of multilayer perceptron have tuned using different optimization techniques. Small to complex architecture of Perceptrons have used to carry out experiments. It was observed that smaller or simpler Neural Network is able to capture the function representation more closely than the more complex one. More complex deep networks with higher capacities are more prone to over fit. Hyper parameters tuning techniques such as dropout and regularization have helped to reduce complexities of model. It is also found that hyper parameter optimization techniques such as Bayesian and Hyperband optimization reduce considerable time and effort. These optimization techniques offer accuracies of 83% and 82% respectively on test dataset.