Support Vector Research Articles

Application of algorithmic models of machine learning to the freight transportation process

The results of the analysis of algorithmic models of machine learning application to the freight transportation process are given in this paper. Analysis of existing research allowed discovering a range of advantages in the application of computational intelligence in logistic systems, including increasing the accuracy of forecasting, reduction of transport costs, increasing the efficiency of cargo delivery, risks reduction, and search for key performance factors. In the research process, the main directions of application of algorithmic models of machine learning were determined. They are vehicle routing, choice of cargo type, transportation type and vehicle type; forecasting fuel consumption by vehicles, disruptions in transportation, transport costs, duration of the order fulfillment; evaluation of the rolling stock fleet and the efficiency of carrying out the transport task. Based on the researched publications, the most common algorithmic models of machine learning in freight transportation were identified, and their effectiveness was analyzed. Linear and logistic regression models are simple enough; however, they do not always provide high simulation results. Deep learning models are quite widely applied to all identified areas. Decision tree and random forest models often show the highest simulation performance. Models of k-nearest neighbors and support vectors should be used both in classification tasks, for example, in choosing the type of cargo and type of transportation, and for forecasting the fuel consumption and the duration of the transport process.

Transient stability assessment of power systems using support vector regressor and convolution neural network

The electric power system is based on a large-scale non-linear system that creates various stability issues. One of the issues is the Transient Stability Assessment, due to multiple hindrances, the power system aims to maintain its synchronism. In the power system, ensuring reliability in planning, control, and monitoring is essential. With the development of new techniques like electric components, power electronics, renewable power generations, etc., to ensure safety and reliability, and improve economic conditions, human interference is complicated. Therefore, a computer-based assessment is needed. Transient stability assessment (TSA) ensures security and provides stable operation in the power system. This paper proposed support-vector machines(SVM)-based support-vector machines Convolutional Neural Networks (CNN) to assist the operation of the power system (SVM-CNN). The novelty of the proposed work is, it minimizes the workload of operational staff and improves efficiency and ability using SVM-based CNN. With the proposed accuracy rate for the training data set was 97.02 %, and the testing data set was 96.83 %. SVM got 96.44 %, and CNN got 96.68 % in the training data set. and the testing data set produces 95.24 % in SVM and 95.74 % in the CNN model.

Multi-source satellite imagery and point of interest data for poverty mapping in East Java, Indonesia: Machine learning and deep learning approaches

This study proposes a novel approach to provide a more granular poverty map in terms of coverage (up to a grid level with the spatial resolution of 1.5 km) with less cost and time to update to support better poverty monitoring. Two poverty estimation model development scenarios using machine learning and deep learning were evaluated in this study. First, the model is constructed using multisource satellite imagery and geospatial point of interest location of economic infrastructures using zonal statistics feature extraction, which is the proposed approach in this study. The specific multisource satellite imagery to be used as poverty indicators are night-time light intensity (NTL) as an approach to economic activity, normalized difference vegetation index (NDVI) as an approach to detect rural areas based on vegetation, built-up index (BUI) as an approach to detect urban areas based on building distribution, normalized difference water index (NDWI) to detect land cover, land surface temperature (LST) as an approach to detecting urban areas based on surface temperature, carbon monoxide (CO), nitrogen dioxide (NO2), and sulfur dioxide (SO2) as an approach to detecting economic activities based on pollution, as well as point of interest (POI) density, and POI distance which show the accessibility of an area. Second, the model is constructed from daytime multiband and nighttime light intensity satellite imagery using the deep learning architecture of Resnet-34 transfer learning feature extraction. In each scenario, we compared the performance of different machine learning (support vector regression/SVR, decision tree regression/DTR, and random forest regression/RFR) algorithms along with deep learning (multiple layer perceptron/MLP and one dimensional-convolutional neural network/CNN-1D) algorithms. Based on the results of the model development evaluation, the CNN-1D model was chosen as the best model in the first scenario, and the Resnet-34 + MLP model was chosen as the best model in the second scenario. The two best models from the first and second scenarios are then used to predict poverty at the grid level with a spatial resolution of 1.5 km. Based on official poverty data, the estimated poverty mapping built using the CNN-1D model in the first scenario was chosen as the best map with the value of root mean squared error (RMSE) 1.95 and adjusted R2 0.84 at the district level. The results of visual identification revealed that high estimates of poverty are typically found in sparsely inhabited areas surrounded by unoccupied land, which is usually an agricultural area. Otherwise, low poverty estimates, on the other hand, are more likely to be found in highly populated regions with convenient access. Our finding is aligned with the official report of Statistics Indonesia, which claims that poverty in rural regions is higher than in urban areas.

WITHDRAWN: Evaluation and performance of satellite-derived bathymetry algorithms in turbid coastal water: A case study of Vengurla rocks

The coastal region bathymetry is increasingly becoming necessary for emerging needs of navigation and development along coasts. Satellite-derived bathymetry (SDB) has been examined as a viable alternative for hydrographic surveys for the past few decades to reduce the data acquisition efforts in coastal regions and augmentation of periodic updating of Electronic Navigational Charts (ENCs). The previous studies have applied SDB algorithms in less complex waters due to the limitations of SDB algorithms in turbid and varying shallow waters. This paper analyses three different medium resolution satellite imagery data to derive bathymetry in navigably very complex and highly turbid region Vengurla Rocks, situated at the west coast of India. The objective of the study was to evaluate the best suitable technique for SDB in turbid water. The bathymetry product images have been derived using the two most commonly utilized Log Ratio, and Linear Ratio transformation; ENVI 5.3 semi-automated tool using Ratio transform; and three machine learning (ML) regression algorithms using Linear Regressor, Random Forest (RF) Regressor and Support Vector (SVR) Regressor. Among the applied transform and algorithms, the RF algorithm using OLI/Landsat-8 data performed the best, resulting in R 2 of 0.88, RMSE of 2.78 m, and MAE 1.83 m. This work established that open source images of sensor OLI/Landsat-8 satellite provide the best results of SDB estimation in complex turbid water by applying Machine Learning algorithms. However, extreme turbid and complex regions resulted in more erroneous SDB estimation specifying the need for refining algorithms using bio-optical parameters. • Satellite derived bathymetry in complex turbid water. • Challenges in retrieval of water depth. • Depth estimation using machine learning algorithms. • The comparison of Machine Learning algorithms in Satellite-Derived Bathymetry.

Hybrid Approach to Identifying Druglikeness Leading Compounds against COVID-19 3CL Protease.

SARS-CoV-2 is a positive single-strand RNA-based macromolecule that has caused the death of more than 6.3 million people since June 2022. Moreover, by disturbing global supply chains through lockdowns, the virus has indirectly caused devastating damage to the global economy. It is vital to design and develop drugs for this virus and its various variants. In this paper, we developed an in silico study-based hybrid framework to repurpose existing therapeutic agents in finding drug-like bioactive molecules that would cure COVID-19. In the first step, a total of 133 drug-likeness bioactive molecules are retrieved from the ChEMBL database against SARS coronavirus 3CL Protease. Based on the standard IC50, the dataset is divided into three classes: active, inactive, and intermediate. Our comparative analysis demonstrated that the proposed Extra Tree Regressor (ETR)-based QSAR model has improved prediction results related to the bioactivity of chemical compounds as compared to Gradient Boosting-, XGBoost-, Support Vector-, Decision Tree-, and Random Forest-based regressor models. ADMET analysis is carried out to identify thirteen bioactive molecules with the ChEMBL IDs 187460, 190743, 222234, 222628, 222735, 222769, 222840, 222893, 225515, 358279, 363535, 365134, and 426898. These molecules are highly suitable drug candidates for SARS-CoV-2 3CL Protease. In the next step, the efficacy of the bioactive molecules is computed in terms of binding affinity using molecular docking, and then six bioactive molecules are shortlisted, with the ChEMBL IDs 187460, 222769, 225515, 358279, 363535, and 365134. These molecules can be suitable drug candidates for SARS-CoV-2. It is anticipated that the pharmacologist and/or drug manufacturer would further investigate these six molecules to find suitable drug candidates for SARS-CoV-2. They can adopt these promising compounds for their downstream drug development stages.

Smart Home-Based Complex Interwoven Activities for Cognitive Health Assessment

With the prevalence of cognitive diseases, the health industry is facing newer challenges since cognitive health deteriorates gradually over time, and clear signs and symptoms appear when it is too late. Smart homes and the IoT (Internet of Things) have given hope to the health industry to monitor and manage the elderly and the less-abled in the comfort of their homes. Smart homes have been most influential in detecting and managing cognitive diseases like dementia. They can give a comprehensive view of the ADL (Activities of Daily Living) of dementia patients. ADLs are categorized as activities of daily life and complex interwoven activities. First signs of cognitive decline appear when a cognitively impaired individual tries to perform complex activities involving planning, analyzing, calculating, and decision making. Therefore, we analyze individuals’ performance while performing complex activities as opposed to Simple ADL. Artificial Intelligence has been one of health-care’s most promising techniques for prediction and diagnosis. When applied to ADL data, machine learning and deep learning algorithms can conveniently and accurately analyze activity patterns and predict the first signs of cognitive decline. Our proposed work uses machine and deep learning classifiers to classify dementia and healthy individuals by analyzing complex interwoven activity data. We use the subset of the CASAS (Centre of Advanced Studies in Adaptive Systems) dataset for eight complex activities performed by 179 individuals in a smart home setting. decision tree, Naive Bayes, support vector, multilayer perceptron classifiers, and deep neural networks have been used for classification. Their results and performances are compared to determine the best classifier. It is observed that deep neural networks and multilayer perceptron show the best results for classifying dementia vs. healthy individuals when evaluating their complex interwoven activities.

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.

A comparative study of driver torque demand prediction methods

AbstractThe performances of energy management systems or electric vehicles and hybrid electric vehicles are highly dependent on the forecast of future driver torque/power request sequence that affects vehicle efficiency and economy. Since the behaviour of the driver is challenging to model/predict by first‐principles models, modern artificial intelligence algorithms would represent feasible methods for approaching this problem in real‐world automotive systems. This work provides a comparative study and analysis of performances of different data‐driven torque prediction strategies. The studied and compared torque demand prediction techniques are exponentially varying model, linear regression, shallow and deep neural networks, and least square support vector machine‐based approaches. The prediction performance and computational cost of these techniques are evaluated and reported, and the possibility of exploiting these techniques in real‐world scenarios is also discussed.

Evolving Hybrid Partial Genetic Algorithm Classification Model for Cost-effective Frailty Screening: Investigative Study.

BackgroundA commonly used method for measuring frailty is the accumulation of deficits expressed as a frailty index (FI). FIs can be readily adapted to many databases, as the parameters to use are not prescribed but rather reflect a subset of extracted features (variables). Unfortunately, the structure of many databases does not permit the direct extraction of a suitable subset, requiring additional effort to determine and verify the value of features for each record and thus significantly increasing cost.ObjectiveOur objective is to describe how an artificial intelligence (AI) optimization technique called partial genetic algorithms can be used to refine the subset of features used to calculate an FI and favor features that have the least cost of acquisition.MethodsThis is a secondary analysis of a residential care database compiled from 10 facilities in Queensland, Australia. The database is comprised of routinely collected administrative data and unstructured patient notes for 592 residents aged 75 years and over. The primary study derived an electronic frailty index (eFI) calculated from 36 suitable features. We then structurally modified a genetic algorithm to find an optimal predictor of the calculated eFI (0.21 threshold) from 2 sets of features. Partial genetic algorithms were used to optimize 4 underlying classification models: logistic regression, decision trees, random forest, and support vector machines.ResultsAmong the underlying models, logistic regression was found to produce the best models in almost all scenarios and feature set sizes. The best models were built using all the low-cost features and as few as 10 high-cost features, and they performed well enough (sensitivity 89%, specificity 87%) to be considered candidates for a low-cost frailty screening test.ConclusionsIn this study, a systematic approach for selecting an optimal set of features with a low cost of acquisition and performance comparable to the eFI for detecting frailty was demonstrated on an aged care database. Partial genetic algorithms have proven useful in offering a trade-off between cost and accuracy to systematically identify frailty.

Support Vector Machine Classifier via L0/1 Soft-Margin Loss.

Support vector machines (SVM) have drawn wide attention for the last two decades due to its extensive applications, so a vast body of work has developed optimization algorithms to solve SVM with various soft-margin losses. To distinguish all, in this paper, we aim at solving an ideal soft-margin loss SVM: L0/1 soft-margin loss SVM (dubbed as L0/1-SVM). Many of the existing (non)convex soft-margin losses can be viewed as one of the surrogates of the L0/1 soft-margin loss. Despite its discrete nature, we manage to establish the optimality theory for the L0/1-SVM including the existence of the optimal solutions, the relationship between them and P-stationary points. These not only enable us to deliver a rigorous definition of L0/1 support vectors but also allow us to define a working set. Integrating such a working set, a fast alternating direction method of multipliers is then proposed with its limit point being a locally optimal solution to the L0/1-SVM. Finally, numerical experiments demonstrate that our proposed method outperforms some leading classification solvers from SVM communities, in terms of faster computational speed and a fewer number of support vectors. The bigger the data size is, the more evident its advantage appears.

Relay Protection and Automation Algorithms of Electrical Networks Based on Simulation and Machine Learning Methods

The tendencies and perspective directions of development of modern digital devices of relay protection and automation (RPA) are considered. One of the promising ways to develop protection and control systems is the development of fundamentally new algorithms for recognizing emergency modes. They work in accordance with the triggering rule, which is formed after processing the results of model experiments. These algorithms are able to simultaneously control a large number of features or mode parameters (current, voltage, resistance, phase, etc.). Thus, the algorithms are multidimensional. This approach in RPA becomes available since the computing power of modern processors is quite enough to process the required amount of statistical data on the parameters of possible normal and emergency operation modes of electrical network sections. The application of classical machine learning algorithms in RPA tasks is analyzed, in particular, methods of k-nearest neighbors, logistic regression, and support vectors. The use of specialized trainable triggering elements is studied both for building new protections and for improving the sophistication of traditional types of relay protection devices. The developed triggering elements of the multi-parameter RPA contribute to an increase in the sensitivity and recognition of accidents. The proposed methods for recognizing emergency modes are appropriate for implementation in intelligent electronic devices (IEDs) of digital substations.

Estimation of shear sonic logs in the heterogeneous and fractured Lower Cretaceous of the Danish North Sea using supervised learning

AbstractShear wave velocity information is valuable in many aspects of seismic exploration and characterization of reservoirs. However, shear wave logs are not always available in the interval of interest due to cost and time‐saving purposes. In this study, we present a tailored supervised learning approach to estimate shear wave velocity from well‐log measurements in the Lower Cretaceous succession of the Valdemar and Boje fields in the Danish North Sea. Our objective is to investigate the performance of four supervised learning regression models (linear, random forest, support vector and multi‐layer perceptron). A limited well‐log data set from six wells is used for training and testing the supervised learning models. A set of well data containing normalized gamma ray, compressional wave velocity, neutron porosity and medium resistivity logs gave reasonable shear wave velocity estimates in the test wells with root‐mean‐square error scores within the range of other published studies. Based on limited input data and complex geology, the multi‐layer perceptron was the most successful model in predicting the reservoir sections of the test wells. However, all models lacked stability in the overburden zones. Lastly, re‐training the multi‐layer perceptron on the six wells to predict missing shear wave velocity in a nearby well showed promising results for further reservoir characterization. The obtained results can yield useful input into, for example, seismic pre‐stack inversion, amplitude versus offset analysis and rock physics analysis.

Evaluation of Mutton Adulteration under the Effect of Mutton Flavour Essence Using Hyperspectral Imaging Combined with Machine Learning and Sparrow Search Algorithm.

The evaluation of mutton adulteration faces new challenges because of mutton flavour essence, which achieves a similar flavour between the adulterant and mutton. Hence, methods for classifying and quantifying the adulterated mutton under the effect of mutton flavour essence, based on near-infrared hyperspectral imaging (NIR-HSI, 1000–2500 nm) combined with machine learning (ML) and sparrow search algorithm (SSA), were proposed in this study. After spectral preprocessing via first derivative combined with multiple scattering correction (1D + MSC), classification and quantification models were established using back propagation neural network (BP), extreme learning machine (ELM) and support vector machine/regression (SVM/SVR). SSA was further used to explore the global optimal parameters of these models. Results showed that the performance of models improves after optimisation via the SSA. SSA-SVM achieved the optimal discrimination result, with an accuracy of 99.79% in the prediction set; SSA-SVR achieved the optimal prediction result, with an RP2 of 0.9304 and an RMSEP of 0.0458 g·g−1. Hence, NIR-HSI combined with ML and SSA is feasible for classification and quantification of mutton adulteration under the effect of mutton flavour essence. This study can provide a theoretical and practical reference for the evaluation and supervision of food quality under complex conditions.

Identification of MiRNA-Disease Associations Based on Information of Multi-Module and Meta-Path.

Cumulative research reveals that microRNAs (miRNAs) are involved in many critical biological processes including cell proliferation, differentiation and apoptosis. It is of great significance to figure out the associations between miRNAs and human diseases that are the basis for finding biomarkers for diagnosis and targets for treatment. To overcome the time-consuming and labor-intensive problems faced by traditional experiments, a computational method was developed to identify potential associations between miRNAs and diseases based on the graph attention network (GAT) with different meta-path mode and support vector (SVM). Firstly, we constructed a multi-module heterogeneous network based on the meta-path and learned the latent features of different modules by GAT. Secondly, we found the average of the latent features with weight to obtain a final node representation. Finally, we characterized miRNA–disease-association pairs with the node representation and trained an SVM to recognize potential associations. Based on the five-fold cross-validation and benchmark datasets, the proposed method achieved an area under the precision–recall curve (AUPR) of 0.9379 and an area under the receiver–operating characteristic curve (AUC) of 0.9472. The results demonstrate that our method has an outstanding practical application performance and can provide a reference for the discovery of new biomarkers and therapeutic targets.

Wearable Inertial Sensor-Based Limb Lameness Detection and Pose Estimation for Horses

Accurate objective, automated limb lameness detection and pose estimation play an important role for animal well-being and precision livestock farming. We present a wearable sensor-based limb lameness detection and pose estimation for horse walk and trot locomotion. The gait event and lameness detection are first built on a recurrent neural network (RNN) with long short-term memory (LSTM) cells. Its outcomes are used in the limb pose estimation. A learned low-dimensional motion manifold is parameterized by a phase variable with a Gaussian process dynamic model. We compare the RNN-LSTM-based lameness detection method with a feature-based multi-layer classifier (MLC) and a multi-class classifier (MCC) that are built on support vector machine/K-nearest-neighbors and deep convolutional neural network methods, respectively. Experimental results show that using only accelerometer measurements, the RNN-LSTM-based approach achieves 95% lameness detection accuracy and also outperforms the feature-based MLC or MCC in terms of several assessment criteria. The pose estimation scheme can predict the 24 limb joint angles in the sagittal plane with average errors less than 5 and 10 degs under normal and induced lameness conditions, respectively. The presented work demonstrate the successful use of machine learning techniques for high performance lameness detection and pose estimation in equine science. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Note to Practitioners</i> —Automation technologies are increasingly used for precision agriculture but few have focused on monitoring individual animals in open field for precision livestock farming. Limb lameness detection and pose estimation in open field is labor-intensive, unsafe for farmers, and inefficient. The presented machine learning-enabled, wearable inertial sensor-based design provides an effective and efficient approach for horse limb lameness detection and pose estimation applications. We present an RNN-LSTM for lameness detection and an integrated manifold learning model is used to predict the horse limb joint angles in walk and trot gaits under normal and induced lameness conditions. We also present a systematic analysis and experiments to demonstrate the impacts of the wearable sensor locations and signal information on lameness detection and pose estimation performance. Several other machine learning-based lameness detection methods are also presented and compared. The extensive multi-horse testing results are presented to demonstrate the superior accuracy and higher performance than other types of machine learning methods. One attractive feature of the proposed design lies in its high performance and fast computational capability for potential real-time applications in open field.

Instructor-learner body coupling reflects instruction and learning

It is widely accepted that nonverbal communication is crucial for learning, but the exact functions of interpersonal coordination between instructors and learners remain unclear. Specifically, it is unknown what role instructional approaches play in the coupling of physical motion between instructors and learners, and crucially, how such instruction-mediated Body-to-Body Coupling (BtBC) might affect learning. We used a video-based, computer-vision Motion Energy Analysis (MEA) to quantify BtBC between learners and instructors who used two different instructional approaches to teach psychological concepts. BtBC was significantly greater when the instructor employed a scaffolding approach than when an explanation approach was used. The importance of the instructional approach was further underscored by the fact that an increase in motion in the instructor was associated with boosted BtBC, but only during scaffolding; no such relationship between the instructor movements and BtBC was found during explanation interactions. Finally, leveraging machine learning approaches (i.e., support vector and logistic regression models), we demonstrated that both learning outcome and instructional approaches could be decoded based on BtBC. Collectively, these results show that the real-time interaction of teaching and learning bodies is important for learning and that the instructional approach matters, with possible implications for both in-person and online learning.

Capacitive voltage transformer measurement error prediction by improved long short-term memory neural network

This paper proposes an improved Long Short-Term Memory neural network (LSTM) for Capacitor Voltage Transformer (CVT) measurement error prediction. The proposed model introduces bidirectional memory, deep feature extraction, and multi-task learning strategies to improve LSTM for high accuracy and high convergence speed. Then, the network parameters are optimized from the multi-source heterogeneous data of CVT, which combines the advantages of prior knowledge and intelligent prediction method. Finally, the trained model is utilized to predict the measurement error automatically. Ablation experiment results show that the three improved strategies can increase prediction accuracy. Comparative experiments show that the proposed improved LSTM model is more accurate than the mainstream machine learning algorithms, such as support vector, random forest, and entropy weight combination forecasting. The proposed improved LSTM model can achieve the CVT ratio error and angle error prediction more accurately and more rapidly than mainstream deep learning algorithms, such as convolutional neural networks, recurrent neural networks, and LSTM.

Improved Boundary Support Vector Clustering with Self-Adaption Support

Concerning the good description of arbitrarily shaped clusters, collecting accurate support vectors (SVs) is critical yet resource-consuming for support vector clustering (SVC). Even though SVs can be extracted from the boundaries for efficiency, boundary patterns with too much noise and inappropriate parameter settings, such as the kernel width, also confuse the connectivity analysis. Thus, we propose an improved boundary SVC (IBSVC) with self-adaption support for reasonable boundaries and comfortable parameters. The first self-adaption is in the movable edge selection (MES). By introducing a divide-and-conquer strategy with the k-means++ support, it collects local, informative, and reasonable edges for the minimal hypersphere construction while rejecting pseudo-borders and outliers. Rather than the execution of model learning with repetitive training and evaluation, we fuse the second self-adaption with the flexible parameter selection (FPS) for direct model construction. FPS automatically selects the kernel width to meet a conformity constraint, which is defined by measuring the difference between the data description drawn by the model and the actual pattern. Finally, IBSVC adopts a convex decomposition-based strategy to finish cluster checking and labeling even though there is no prior knowledge of the cluster number. Theoretical analysis and experimental results confirm that IBSVC can discover clusters with high computational efficiency and applicability.

CARD9: Cardiac Contractility Estimation Based on Left Ventricular Pressure: Comparison of Time Series Classifiers and Graph-based Neural Networks

Background: In heart failure therapy, the real-time estimation of remaining cardiac contractility (RCC) can facilitate patient management, as well as, the performance of physiological controllers for ventricular assist devices (VAD). In this study, the real-time estimation of RCC based on time series data (TSD) of the left ventricular pressure (LVP) was investigated by exploiting two traditional time series classifiers (TSC) and two graph-based neural networks (GNN). All classifiers were assessed with respect to their RCC estimation accuracy and their applicability for real-time VAD control. Methods: For the RCC estimation (Figure 1), LVP TSD was generated using a hybrid mock circulation (HMC). On the HMC instantaneous pressure, volume, and flowrate values were computed by a numerical model of the human cardiovascular system and applied by a hydraulic interface on the inlet and outlet mixed-flow blood pump. The experimental protocol for the acquisition of the LVP TSD included 63 experiments, where preload, afterload, pump speed, and heart rate variations were performed for 9 RCC values. The data was pre-processed by segmenting and normalizing the TSD, producing 6300 cardiac cycle samples, which were used for the TSC approaches. The TSCs comprised the dynamic time warping nearest neighbor (DTW-NN) classifier and the support vector (SVM) classifier. Furthermore, the maturity of two GNN approaches was explored, including a simple custom architecture (C-GNN) and a pretrained architecture (P-GNN) provided by the KERAS library (Inception V3). The C-GNN comprised a convolutional layer with 32 filters of size 3x3, a max pooling layer with filter size 2x2, a dense layer with 128 units, and a dense layer with 9 units for the classification output. For the TSD to be used in the GNN frameworks, two different image encoders (IE), namely direct plot and recurrence plot encoders, were used and assessed. The data was split 80/20 for network training and validation. Results: All four classifiers, DTW-NN, SVM, C-GNN (IE: recurrence plot), and the P-GNN (IE: direct plot) achieved an accuracy of at least 98%. The SVM had the highest accuracy (99.9%) and the shortest prediction time (63 μs/sample). The prediction time of DTW-NN was on average 4.84 s/sample, prohibiting the implementation in real-time VAD control. Both GNN architectures achieve high accuracy and short prediction time, however, their performance does not reach the benchmark set by the SVM. A comparison of the image encoders showed that both the direct plot and the recurrence plot encoder lead to similar classification performances when using the same GNN, but different performances between the two investigated GNNs. Conclusion: All classification methods provided accurate RCC estimation with the SVM showing superior performance and being the most promising for real-time clinical implementation. These estimation approaches could substantially support patient surveillance and physiologic VAD control approaches. Figure 1. Schematic of the sequential steps for the estimation of the remaining cardiac contractility from time-series data (TSD) of the left ventricular pressure (LVP). a) LVP TSD is produced through a hybrid mock circulation, where the human cardiovascular system is simulated and the pressures, volume, and flowrates calculated are applied on the inlet and outlet of a ventricular assist device in real-time via a hydraulic interface. LVP TSD was acquired for various physiologic conditions. b) Pre-processing of the data, including cardiac cycle segmentation and time normalization. c) Translation of the TSD to images through different image encoders. d) Time series classifiers, including dynamic time warping nearest neighbor (DTW-NN) and the support vector (SVM) classifiers. e) Graph-based neural networks (GNN). f) Classified cardiac contractility.

Development of a novel compressive strength design equation for natural and recycled aggregate concrete through advanced computational modeling

Owing to the variations in the recycled coarse aggregates (RCA) characteristics, the compressive strength prediction of recycled aggregate concrete (RAC) is a complex challenge causing hindrance in the design guidelines development and practical application of RAC. This study aims to develop a unified compressive strength model for the RAC and natural aggregate concrete (NAC) independent of RCA source and other properties. For this reason, four input parameters, including water absorption of coarse aggregates, effective water-to-cement ratio, coarse aggregates to cement ratio, and RCA replacement ratio, are considered to predict the compressive strength of NAC and RAC. Ten machine-learning techniques, including random forest, gradient boost, Ada boost, k-nearest neighbor, bagging regressor, support vector, XG boost, decision tree, artificial neural network, and gene expression programming, are evaluated through a test database having 962 experimental results of compressive strength of NAC and RAC from 107 different studies. The performance of machine-learning algorithms is assessed through various statistical parameters. Results show that the input parameters (considered in this study) are essential in predicting the cubic and cylindrical compressive strength of NAC and RAC. The machine learning models and comprehensive design equations developed in this study are better than existing models and can be recommended as an effective tool for predicting the compressive strength of NAC and RAC having RCA from different sources, leading toward the development of sustainable concrete design guidelines.