Efficient Machine Learning Model Research Articles

Efficient Machine Learning Models for the Uplift Behavior of Helical Anchors in Dense Sand for Wind Energy Harvesting

Helical anchors are widely used in engineering to resist tension, especially during offshore wind energy harvesting, and their uplift behavior in sand is influenced by many factors. Experimental studies are often used to investigate these anchors; however, scale effects are inevitable in 1× g model tests, soil conditions for in situ tests are challenging to control, and centrifuge tests are expensive and rare. To make full use of the limited valid data and to gain more knowledge about the uplift behaviors of helical anchors in sand, a prediction model integrating gradient-boosting decision trees (GBDT) and particle swarm optimization (PSO) was proposed in this study. Data obtained from a series of centrifuge tests formed the dataset of the prediction model. The relative density of soil, embedment ratio, helix spacing ratio, and the number of helices were used as input parameters, while the anchor mobilization distance and the ultimate monotonic uplift resistance were set as output parameters. A GBDT algorithm was used to construct the model, and a PSO algorithm was used for hyperparameter tuning. The results show that the optimal GBDT model accurately predicted the anchor mobilization distance and the ultimate monotonic uplift resistance of helical anchors in dense fine silica sand. By analyzing the relative importance of influencing variables, the embedment ratio was found to be the most significant variable in the model, while the relative density of the fine silica sand soil, the helix spacing ratio, and the number of helices had relatively minor influence. In particular, the helix spacing ratio was found to have no influence on the capacity of adjacent helices when S/D > 6.

Inverse Estimation of Breast Tumor Size and Location with Numerical Thermal Images of Breast Model Using Machine Learning Models

Early screening of cancer plays a vital role in successful treatment. Normally, the temperature of breast surface changes due to the size and location of the underlying tumor. Temperature patterns can be used to estimate the size and location of tumor by an inverse approach using machine learning algorithms. The present study aims to provide an efficient machine learning model that can be relied on to predict the size and location of the tumor using numerical thermal images. There is no availability of actual thermal images labeled with the size and location of the tumor. Consequently, successive numerical simulations are used to develop the numerical thermal image dataset with a three-dimensional breast model solved with COMSOL Multiphysics. The obtained numerical thermal images are uniquely labeled with the corresponding size and location of the tumor and trained using various machine learning regression models such as linear, support vector, K-nearest neighbor, and decision tree regression models. The results are analyzed using parity plots and mean absolute error. The present study found that the decision tree regression model outperformed the other machine learning models and improved the estimation accuracy by rejecting the numerical thermal images which are having slight variation in surface temperature.

Ensemble Machine Learning approach for evaluating the material characterization of carbon nanotube-reinforced cementitious composites

Time and cost-efficient techniques are essential to avoid extra conventional experimental studies with large data-set for material characterization of composite materials. This study is aimed at providing a correlation between the structural performance and mechanical properties of carbon nano-tubes reinforced cementitious composites through efficient predictive Machine Learning (ML) models. The Flexural (FS) and Compressive (CS) Strength of Carbon Nanotube (CNT)-reinforced composites were predicted based on the data-rich framework provided in the literature. Two different ensembled ML methods including Random Forest (RF) and Gradient Boosting Machine (GBM) were implemented on those data for predicting the CNT-reinforced cement-based composites. Data-set were utilized for training the proposed models through employing SciKit-Learn library in Python, followed by hyper-parameter tuning and k-fold cross-validation method for obtaining an optimum model to predict the target values. It was shown that the CS values predicted by the proposed models were more accurate than the FS counterparts and the developed model of GBM has less sensitivity to the alteration of test data than the proposed RF model. Finally, sensitivity analysis was conducted through Sobol algorithm and the parameters with highest contribution were identified.

Machine learning-based inverse design for single-phase high entropy alloys

In this work, we develop an inverse design framework to search for single-phase high entropy alloys (HEAs) subjected to specified phase targets and constraints. This framework is based on the fast grid search in the composition–temperature space, enabled by a highly accurate and efficient machine learning model trained by a huge amount of data. Using the framework, we search through the entire quaternary, quinary, and senary alloy systems, formed by Al, Co, Cr, Cu, Fe, Mn, Ni, and Ti, to identify three types of HEAs: (1) the single-phase FCC HEA with the highest Al content; (2) the single-phase FCC HEA with lower equilibrium temperatures; and (3) single-phase BCC HEAs with Al as the principal element. For the first time, we reveal that the highest Al content in single-phase FCC HEAs is 0.15 in mole fraction, which is higher than the Al contents in all reported single-phase FCC HEAs. The identified HEAs for the quaternary, quinary, and senary groups are Al0.15Co0.34Cr0.16Ni0.35, Al0.15Co0.35Cr0.1Fe0.05Ni0.35, and Al0.15Co0.36Cr0.06Fe0.06Mn0.01Ni0.36, respectively. All the designed HEAs are verified by the equilibrium calculations with Thermo-Calc software and the TCHEA3 database. We further conduct Scheil–Gulliver calculations and experimental fabrications and characterizations for the designed HEAs, to verify the formation of the targeted phases at non-equilibrium conditions. This work demonstrates a viable approach to design HEAs with specified phase targets and constraints.

Deciphering the importance of MD descriptors in designing Vitamin D Receptor agonists and antagonists using machine learning

The Vitamin D Receptor (VDR) ligand-binding domain undergoes conformation change upon the binding of VDR agonists/antagonists. Helix 12 ((H)12) is one of the important helices at VDR ligand binding and its conformational changes are controlled by the binding of agonists and antagonists molecules. Various molecular modeling studies are available to explain the agonistic and antagonistic activity of vitamin D analogs. In this work, for the first time, we attempted to generate a machine learning model with fingerprints, 2D, 3D and MD descriptors that are specific to Vitamin D analogs and VDR. Initially, 2D and 3D descriptors and fingerprints of 1003 vitamin D analogs were calculated using CDK and RDKit. The machine learning model was generated using descriptors and fingerprints. Further, 80 Vitamin D analogs (40 VDR agonists + 40 VDR antagonists) were docked in the VDR active site. 50ns MD simulation was performed for each protein-ligand complex. Different MD descriptors such as Solvent Accessible Surface Area (SASA), radius of gyration, PC1 and PC2 were calculated and considered along with CDK and RDKit descriptors as features for machine learning calculations. A few other descriptors that are related to VDR conformational changes such as conformation of the (H)12, the angle at kink were considered for machine learning model generation. It was observed that the descriptors calculated from VDR conformational changes i) were able to distinguish between agonists and antagonists ii) provide key and comprehensive information about the unique binding characteristics of agonists and antagonists iii) provide a strong basis for the machine learning model generation. Overall, this study attempts the utilization of descriptors that are specific to a protein conformation will be helpful for the generation of an efficient machine learning model.

A Machine Learning Model for Estimation of Village Level Soil Nutrient Index

Objectives: To propose an innovative technique for designing an efficient and adaptive machine learning model using classifier assembly for estimating village level soil nutrient index using soil datasets. Methods: Freely available soil datasets were collected from the concerned authority of Govt. of India. These datasets were used by the proposed machine learning model designed with a classifier assembly of fifteen diverse classifiers for nutrient class identification. The performance of each classifier was evaluated in terms of five well-accepted standard metrics. The outputs of the best performing classifier were then used for estimation of village level nutrient index using modified Parker’s method. Findings: The model was applied for nutrient class identification, and estimation of the nutrient index of different villages using freely available benchmarked Soil health Card datasets provided by the Govt. of India. The empirical results depicted that for nutrient class identification, this proposed machine learning model overperformed the other existing models in terms of average accuracy score. In the case of Copper, it provided the highest average accuracy of classification (0.949) and estimation accuracy of 95.48%. For Sulphur, an average classification accuracy of 0.891 and an estimation accuracy of 90.66% were achieved. Similarly, for Zinc, an average classification accuracy of 0.883 and an estimation accuracy of 89.63% were observed. Novelty: This study suggests a novel architecture of a machine learning model using classifier assembly to estimate the village level nutrient index with the highest possible accuracy, using freely available soil datasets. Keywords: Nutrient index; Village level soil fertility; Fertilizer management; Machine learning; Classifier assembly

Optimization of Proton Exchange Membrane Electrolyzer Cell Design Using Machine Learning

We propose efficient multiple machine learning (ML) models using specifically polynomial and logistic regression ML methods to predict the optimal design of proton exchange membrane (PEM) electrolyzer cells. The models predict eleven different parameters of the cell components for four different input parameters such as hydrogen production rate, cathode area, anode area, and the type of cell design (e.g., single or bipolar). The models fit well as we trained multiple machine learning models on 148 samples and validated the model performance on a test set of 16 samples. The average accuracy of the classification model and the mean absolute error is 83.6% and 6.825, respectively, which indicates that the proposed technique performs very well. We also measured the hydrogen production rate using a custom-made PEM electrolyzer cell fabricated based on the predicted parameters and compared it to the simulation result. Both results are in excellent agreement and within a negligible experimental uncertainty (i.e., a mean absolute error of 0.615). Finally, optimal PEM electrolyzer cells for commercial-scaled hydrogen production rates ranging from 500 to 5000 mL/min were designed using the machine learning models. To the best of our knowledge, we are the first group to model the PEM design problem with such large parameter predictions using machine learning with those specific input parameters. This study opens the route for providing a form of technology that can greatly save the cost and time required to develop water electrolyzer cells for future hydrogen production.

Machine learning prediction of specific capacitance in biomass derived carbon materials: Effects of activation and biochar characteristics

The preparation process of biomass-based biochar materials is usually screened using traditional trial-and-error experiments. In this approach, the electrochemical properties of biochar are correlated with properties called descriptors. In this work, several simple and efficient machine learning (ML) models were used to predict the electrical capacity of biochar through activation conditions, biochar properties, and testing conditions. The established ML model predicted the capacitance of biochar with 9 descriptors that are readily available values during the preparation of biochar. The prediction performance of four regression methods (Decision Tree (DT), Artificial Neural Network (ANN), eXtreme Gradient Boosting (XGBoost) and Random Forest (RF)) were evaluated with a test set/training set ratio of 8 to 2. Among the four regression methods, XGBoost had the best prediction effect on the electrochemical performance of biochar with a low mean root mean square error (RMSE) and coefficient of determination (R2) close to 1. In addition, the analysis of the importance of the features under each model combined with the existing research verifies the rationality of the model. The accuracy and simplicity of this system demonstrate that the electrochemical performance of biochar can be easily predicted without time-consuming traditional experimental procedures and can be a method to guide the direction of experiments.

Protection of a smart grid with the detection of cyber- malware attacks using efficient and novel machine learning models

False data injection (FDI) attacks commonly target smart grids. Using the tools that are now available for detecting incorrect data, it is not possible to identify FDI attacks. One way that can be used to identify FDI attacks is machine learning. The purpose of this study is to analyse each of the six supervised learning (SVM-FS) hybrid techniques using the six different boosting and feature selection (FS) methodologies. A dataset from the smart grid is utilised in the process of determining the applicability of various technologies. Comparisons of detection strategies are made based on how accurately each one can identify different kinds of threats. The performance of classification algorithms that are used to detect FDI assaults is improved by the application of supervised learning and hybrid methods in a simulated exercise.

A new feature extraction approach of medical image based on data distribution skew

Building a highly efficient machine learning model requires sufficient data to allow robust feature extraction capable of recognizing patterns in each class; thus, the model can distinguish among different classes. It is important to extract effective features from the available amount of data without the need for more real data or improve them using an augmentation technique. The matter gets more complicated if the data is of the image type. In this paper, a new approach for feature extraction called Feature Extraction Based on Region of Mines (FE_mines) is presented that includes three versions to deal with different medical images; this approach obtains multiple formulas for each image using the signal and image processing, then data distribution skew is used to calculate three statistical measurements that include the hidden features, which leads to increased discrimination among classes to build powerful models with better performance and high efficiency. Three experiments were conducted using three types of medical image datasets, namely: Diabetic Retinopathy (Color Fundus photography); Brain Tumor (MRI); and COVID-19 chest (X-ray). The results proved that the FE_mines approach achieved higher accuracy ranges (1 to 13)% within the three experiments than the two traditional methods (RGB and ASPS approaches). In addition, an augmentation technique to increase the size of the dataset is not required which has negative effects on performance. Furthermore, the approach simultaneously included three preprocessing techniques: feature selection, reduction, and extraction.

Simulation-based optimization of large-scale dedicated bus lanes allocation: Using efficient machine learning models as surrogates

Dedicated Bus Lanes (DBLs) have been implemented in many cities to boost buses’ reliability and to alleviate traffic congestions. However, how to correctly allocate DBLs across a large-scale real-world network is challenging. Simulation-Based Optimization (SBO) methods were utilized in this work to resolve this optimal allocation problem. Traditional discrete SBO methods are intractable when handling high-dimensional, costly, simulation-based Transportation Network Design Problems (TNDPs) using a limited computational budget. Thus, several efficient Machine Learning (ML)-based surrogate models and a Jackknife uncertainty estimator were introduced to existing SBO framework in this work. A number of comparative experiments between proposed methods and frequently-used Gaussian Process (GP)-SBO methods were conducted. A mesoscopic simulation and Dynamic Traffic Assignment (DTA) tool was adopted to evaluate the network performance.The results of numerical studies show that the optimization efficiency of proposed method is significantly higher than that of commonly used GP-based method when dealing with high dimensional problems. A real-world DBLs allocation case study in Guiyang, Guizhou Province, China again proves that efficient ML-based SBO method is capable to take much less CPU runtime to obtain a better solution than traditional method. The optimal DBLs allocation scheme found by one of the efficient ML-based methods raises the network performance by 5.05 %. A total of 1,376 h travel time is saved, and the average travel time per traveler drops by 0.75 min. In conclusion, efficient ML-based SBO method proposed in this study is more promising to handle large-scale, discrete, costly simulation-based DBLs allocation problems within a limited computational budget than common GP-based SBO methods.

Application of machine learning methods for estimating and comparing the sulfur dioxide absorption capacity of a variety of deep eutectic solvents

Sulfur dioxide (SO 2 ) is one of the main atmospheric pollutants and an active threat to human health. SO2 separation from industrial flue gases improves air quality, decreases human health problems, and reserves sulfur resources. The capturing and recycling processes demand a green solvent with an effective, selective, and reversible SO 2 absorption. Deep eutectic solvents (DESs) have recently been engaged in SO 2 capture/recycle processes. The SO 2 removal capacity of DESs depends on their ingredients (the hydrogen bond donor, hydrogen bond acceptor, water content), temperature, and pressure. Despite comprehensive experimental investigations, literature presents no clue to compare the SO 2 absorption capacity of DESs considering their compositions and operating conditions. Therefore, this work deploys an efficient machine learning model to estimate the SO 2 absorption capacity of DESs as a function of their molecular weight, water content, pressure, and temperature. All the laboratory-measured datasets reported in the literature have been included to ensure that the deployed model is reliable and generalized. Furthermore, the proposed model has been selected among five different classes of artificial neural networks . A single hidden-layer neural network with only eleven neurons optimized by the Levenberg-Marquardt is the most precise model predicting 480 DES-SO 2 phase equilibria with the mean squared error , mean absolute percentage error, and coefficient of determination of 1.13 × 10 −3 , 4.76%, and 0.97936, respectively. This research is the first step toward constructing a reliable model for screening available DESs based on their SO 2 absorption capacity and finding the best candidate. • Two scenarios have been followed for estimating the SO 2 solubility in 29 DESs. • Various machine learning techniques were designed/compared to find the best model. • Systematic ranking analyses confirm that the MLP-LM is the highest accurate model. • The model shows excellent performance, i.e., R 2 = 0.97936, MSE = 1.13 × 10 −3 , MAPE = 4.76%. • Trend analyses confirm that the MLP-LM has compatibility with the system behavior.

Precise prediction of multiple anticancer drug efficacy using multi target regression and support vector regression analysis

Background and objectivesThe prediction of multiple drug efficacies using machine learning prediction techniques based on clinical and molecular attributes of tumors is a new approach in the field of precision medicine of oncology. The selection of suitable and effective therapeutic drugs among the potential drugs is performed computationally considering the tumor features. In this study, we developed and validated machine learning models to predict the efficacy of five anti-cancer drugs according to the clinical and molecular attributes of 30 oral squamous cell carcinoma (OSCC) cohorts. This sounds a bit odd – consider: Ranking of the drugs was achieved using their apoptotic priming. MethodsWe developed multiple drug efficacy prediction models based on three types of tumor characteristics by applying machine learning methods, including multi-target regression (MTR) and support vector regression (SVR). The prediction accuracy of existing machine learning methods was enhanced by introducing novel pre-processing techniques to develop Enhanced MTR (E_MTR), Enhanced Log-based MTR (EL_MTR), Enhanced Multi-target SVR (EM_SVR), and Enhanced Log-based Multi-target SVR (ELM_SVR). As a unique capability, ELM_SVR and EL_MTR rank the drugs based on their predicted efficacy. All the drug efficacy prediction models were built using OSCC real samples and theoretical samples. The best model was selected was based on dataset size and evaluation metrics, such as error terms, residuals and parameter tuning, and cross-validated (CV) using 30 real samples and 340 theoretical samples. ResultsWhen 30 real tumor samples were used for the train-test and CV methods, MTR models predicted the efficacy with less error than SVR models. Comparatively, using 340 theoretical samples for the train-test and CV methods, though MTR improved the performance, SVR predicted the efficacy with zero error. We found that, for small samples, the proposed MTR provided a 0.01 difference between actual apoptotic priming and predicted priming of five drugs. For large samples, the predicted values by the proposed SVR had a difference of 0.00001. The error terms (Actual vs. Predicted) also reveal that the enhanced log model is suitable when MTR is applied. Meanwhile, the enhanced model is suitable for SVR learning for multiple drug efficacy prediction. It was found that the predicted ranks of the drugs based on the multi-targeted efficacy prediction exactly match the actual rankings. ConclusionWe developed efficient statistical and machine learning models using MTR and SVR analysis for anticancer drug efficacy, which will be useful in the field of precision medicine to choose the most suitable drugs in personalized manner. The performance results of the proposed enhanced ranking techniques are described as follows: i) EL_MTR is the best to predict multiple anticancer drug efficacies and improve the accuracy of ranking drugs, irrespective of sample size; and ii) ELM_SVR performs better than other MTR models with a large sample size and precise ranking process.

An Efficient Machine Learning Model Based on Improved Features Selections for Early and Accurate Heart Disease Predication.

Coronary heart disease has an intense impact on human life. Medical history-based diagnosis of heart disease has been practiced but deemed unreliable. Machine learning algorithms are more reliable and efficient in classifying, e.g., with or without cardiac disease. Heart disease detection must be precise and accurate to prevent human loss. However, previous research studies have several shortcomings, for example,take enough time to compute while other techniques are quick but not accurate. This research study is conducted to address the existing problem and to construct an accurate machine learning model for predicting heart disease. Our model is evaluated based on five feature selection algorithms and performance assessment matrix such as accuracy, precision, recall, F1-score, MCC, and time complexity parameters. The proposed work has been tested on all of the dataset'sfeatures as well as a subset of them. The reduction of features has an impact on theperformance of classifiers in terms of the evaluation matrix and execution time. Experimental results of the support vector machine, K-nearest neighbor, and logistic regression are 97.5%,95 %, and 93% (accuracy) with reduced computation timesof 4.4, 7.3, and 8seconds respectively.

Development of machine learning models for the prediction of laminar flame speeds of hydrocarbon and oxygenated fuels

Laminar flame speed (LFS) is a key physicochemical property of a premixed fuel/oxidizer mixture, and is critical in the description of complex combustion phenomena. Accurate experimental measurements of LFSs for various fuels have been performed to develop and validate detailed kinetic mechanisms, which in turn are used to predict LFSs under various combustion conditions. However, such procedure is inefficient, especially in large-scale turbulent combustion modeling studies. Based on previous experimental studies of LFSs for various fuels, this work aims to develop a data-driven machine learning (ML) model for the prediction of LFSs of hydrocarbon and oxygenated fuels. Descriptors computed from semi-empirical quantum chemistry methods are used as input in ML models due to the simplicity and computational-efficiency. Pearson correlation analysis is used to select important features, and 5 descriptors are screened as the input features for ML model development. The accuracies and interpretabilities of existing 16 ML algorithms in the prediction of LFSs are compared through systematically evaluated the errors based on the differences between experimental data and model prediction. These ML models include regression trees, support vector machine regression, gaussian process regression, and ensemble trees. An efficient ML model for predicting LFSs of hydrocarbon and oxygenated fuels based on gaussian process regression algorithm is proposed, which exhibits good accuracy in predicting of LFSs for variable pressure, temperature, and equivalence ratio. The dependency of LFSs on the descriptors are also analysed. The developed ML model is fast enough for integration into large-scale computational fluid dynamics for combustion studies.

Probing the stochastic fracture behavior of twisted bilayer graphene: Efficient ANN based molecular dynamics simulations for complete probabilistic characterization

The present article outlines a probabilistic investigation of the uniaxial tensile behaviour of twisted bilayer graphene (tBLG) structures. In this regard, the twist angle (θ) and temperature (T) are considered as the control parameters and the ultimate tensile strength (UTS) and failure strain of the tBLG structures are considered as the responses. It is observed that with the increase in twist angle (θ) of tBLG; the fracture responses exhibit a declining trend deterministically. The tBLG twisted with the magic angle (θ = 1.08o) results in around 7% decrease in UTS and nearly 24% decrease in failure strain, when compared with normal BLG (θ = 0̊). The Monte Carlo simulation (MCS) based random sampling is performed for the considered control parameters, wherein θ is varied from 0o to 30o, and temperature is varied from 100 K to 900 K. Within such bounds of the input parameters, the training (64 samples) and validation (8 samples) sample spaces are constructed. In the next step, molecular dynamics (MD) simulation of uniaxial tensile deformation of the modelled tBLG structures is carried out for each instance of the sample space. The dataset is subsequently used to form and validate the artificial neural network (ANN). The computationally efficient machine learning (ML) model is further utilized to perform the detailed investigation of fracture behaviour of the tBLG structures in the probabilistic framework. Such analysis captures all the possible instances of variation in the input parameters and leads to deep insights in the material behaviour, which would have otherwise remained unnoticed due to the prohibitive nature of conducting a large number of MD simulations. The novelty of the present study lies in the probabilistic interpretation of the tensile behaviour of tBLG structures subjected to variation in twist angle (θ) and temperature (T). The preparation of nano-scale samples with the exact design specifications such as twist angle is often extremely difficult, which leads to inevitable stochastic system disorders. The current article essentially proposes a probabilistic avenue of quantifying the effect of such disorders on the failure properties of tBLG.

Multi-fidelity surrogate modeling through hybrid machine learning for biomechanical and finite element analysis of soft tissues

Biomechanical simulation enables medical researchers to study complex mechano-biological conditions, although for soft tissue modeling, it may apply highly nonlinear multi-physics theories commonly implemented by expensive finite element (FE) solvers. This is a significantly time-consuming process on a regular computer and completely inefficient in urgent situations. One remedy is to first generate a dataset of the possible inputs and outputs of the solver in order to then train an efficient machine learning (ML) model, i.e., the supervised ML-based surrogate, replacing the expensive solver to speed up the simulation. But it still requires a large number of expensive numerical samples. In this regard, we propose a hybrid ML (HML) method that uses a reduced-order model defined by the simplification of the complex multi-physics equations to produce a dataset of the low-fidelity (LF) results. The surrogate then has this efficient numerical model and an ML model that should increase the fidelity of its outputs to the level of high-fidelity (HF) results. Based on our empirical tests via a group of diverse training and numerical modeling conditions, the proposed method can improve training convergence for very limited training samples. In particular, while considerable time gains comparing to the HF numerical models are observed, training of the HML models is also significantly more efficient than the purely ML-based surrogates. From this, we conclude that this non-destructive HML implementation may increase the accuracy and efficiency of surrogate modeling of soft tissues with complex multi-physics properties in small data regimes.

Federated Learning Over Multihop Wireless Networks With In-Network Aggregation

Communication limitation at the edge is widely recognized as a major bottleneck for federated learning (FL). Multi-hop wireless networking provides a cost-effective solution to enhance service coverage and spectrum efficiency at the edge, which could facilitate large-scale and efficient machine learning (ML) model aggregation. However, FL over multi-hop wireless networks has rarely been investigated. In this paper, we optimize FL over wireless mesh networks by taking into account the heterogeneity in communication and computing resources at mesh routers and clients. We present a framework that each intermediate router performs <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">in-network</i> model aggregation before sending the data to the next hop, so as to reduce the outgoing data traffic and hence aggregate more models under limited communication resources. To accelerate model training, we formulate our optimization problem by jointly considering model aggregation, routing, and spectrum allocation. Although the problem is a non-convex mixed-integer nonlinear programming, we transform it into a mixed-integer linear programming (MILP), and develop a coarse-grained fixing procedure to solve it efficiently. Simulation results demonstrate the effectiveness of the solution approach, and the superiority of the in-network aggregation scheme over the counterpart without in-network aggregation.

Building an Efficient Intrusion Detection System using Feature Selection and Machine Learning

Abstract: With the increase in internet activity and as the world goes digital as the days go, the risk of exposure to malicious activities also increased rapidly. The intruders/hackers use various methods to gain unauthorized access to one's computer or any other device, Network Intrusion is one of the methods by which intruders attack the network of the user, the user can be an individual or an organization based on the intention/agenda of the attackers. Significant Reasons for intrusion are Hacktivism, Steal Money or Data, and Spying. Due to the internet being a vast place, it is challenging to pinpoint a particular way in which Network Intrusion takes place, therefore a Network Intrusion Detection System needs to be put in place in order to deal with the issues regarding Network Intrusions. There are multiple leaks or data extortion that happened previously and, in this paper, the dataset released based on a leak from KDD99 is used. An improved version of KDD99 (NSL-KDD) is used in this study. NSLKDD datasets have been used for training the Machine Learning Model. Given the number of attributes in the dataset, it was difficult to use all attributes so, feature selection methods were used to get the best attributes to develop an efficient Machine Learning model. In this analysis of Machine Learning algorithms, the algorithms under consideration are Logistic Regression, Support Vector Machine, Decision Tree, Random Forest, and Naive-Bayes. For comparison of the performance of the algorithms metrics like Accuracy Score, Confusion Matrix, and Classification Report were considered to find the best algorithm among them.

Scour depth prediction at bridge piers using metaheuristics-optimized stacking system

This paper proposes a novel artificial intelligence-based approach, called the metaheuristics-optimized stacking system (MOSS), to assist civil engineers in estimating scour depth at bridge piers. MOSS integrates the advantages of hybrid and stacking ensemble schemes by combining a metaheuristic optimization algorithm with efficient machine learning (ML) models. The metaheuristic algorithm simultaneously finds the optimal values of all hyperparameters of constituent ML models in the stacking ensemble to generate the most effective system. The developed MOSS was established by integrating a forensic-based investigation (FBI) algorithm, least squares support vector regression (LSSVR), and radial basis function neural network (RBFNN). The efficiency of the MOSS is verified comprehensively with reference to three case studies of both laboratory data and field data that cover various levels of complexity and types of pier foundations. The performance of MOSS is compared to that of other single ML models, conventional voting ensemble, hybrid models, empirical methods, and mathematical approach. The analytical results of cross-validation reveal that MOSS was the most reliable approach, achieving the best values of all performance evaluation metrics. MOSS exhibited mean absolute percentage errors of 7.127%, 29.195% and 13.131% in the prediction of scour depth using a laboratory dataset, a field dataset, and a complex pier foundations dataset, respectively, and these values are at least 36%, 19% and 41% lower than those obtained using other approaches. The automated predictive analytics revealed the robustness, efficiency, and stability of MOSS. The novelty and contributions of this study include (1) providing a general stacking framework and re-defining the establishment of an ensemble ML model, and (2) developing a highly promising tool that greatly outperforms currently available methods to help civil engineers estimate the scour depth around bridge piers.