Improved Machine Learning based Classification Model for Early Autism Detection

The Sixth International Symposium on Neural Networks (ISNN 2009)

Emotion recognition plays a pivotal role in advancing Human-Computer Interaction (HCI) by enabling systems to understand and respond to human emotional states. Among various modalities, Speech Emotion Recognition (SER) stands out as a non-invasive, cost-effective, and temporally efficient approach for detecting emotions. This study leverages the RAVDESS dataset, which provides high-quality audio recordings for emotion classification. The proposed methodology involves preprocessing audio signals to remove noise, extracting temporal and spectral features in both time and frequency domains, and implementing machine learning models for multi-class emotion classification. The study evaluates and compares the performance of several classification models, including Random Forest (RF), Multilayer Perceptron (MLP), Support Vector Machine (SVM), Convolutional Neural Networks (CNN), and Decision Trees (DT). Experimental results demonstrate promising accuracy levels, highlighting the potential of these machine learning techniques in SER. This research contributes to the integration of SER within Brain-Computer Interfaces (BCI) and other emotion-aware applications, paving the way for enhanced interactive systems. Keywords – Speech Emotion Recognition -Machine Learning - Random Forest -Multilayer Perceptron - Support Vector Machine - Convolutional Neural Networks - Decision Tree - RAVDESS Dataset

Background and aimsThe birth weight of a newborn is a crucial factor that affects their overall health and future well-being. Low birth weight (LBW) is a widespread global issue, which the World Health Organization defines as weighing less than 2,500 g. LBW can have severe negative consequences on an individual’s health, including neonatal mortality and various health concerns throughout their life. To address this problem, this study has been conducted using BDHS 2017–2018 data to uncover important aspects of LBW using a variety of machine learning (ML) approaches and to determine the best feature selection technique and best predictive ML model.MethodsTo pick out the key features, the Boruta algorithm and wrapper method were used. Logistic Regression (LR) used as traditional method and several machine learning classifiers were then used, including, DT (Decision Tree), SVM (Support Vector Machine), NB (Naïve Bayes), RF (Random Forest), XGBoost (eXtreme Gradient Boosting), and AdaBoost (Adaptive Boosting), to determine the best model for predicting LBW. The model’s performance was evaluated based on the specificity, sensitivity, accuracy, F1 score and AUC value.ResultsResult shows, Boruta algorithm identifies eleven significant features including respondent’s age, highest education level, educational attainment, wealth index, age at first birth, weight, height, BMI, age at first sexual intercourse, birth order number, and whether the child is a twin. Incorporating Boruta algorithm’s significant features, the performance of traditional LR and ML methods including DT, SVM, NB, RF, XGBoost, and AB were evaluated where LR, had a specificity, sensitivity, accuracy and F1 score of 0.85, 0.5, 85.15% and 0.915. While the ML methods DT, SVM, NB, RF, XGBoost, and AB model’s respective accuracy values were 85.35%, 85.15%, 84.54%, 81.18%, and 84.41%. Based on the specificity, sensitivity, accuracy, F1 score and AUC, RF (specificity = 0.99, sensitivity = 0.58, accuracy = 85.86%, F1 score = 0.9243, AUC = 0.549) outperformed the other methods. Both the classical (LR) and machine learning (ML) models’ performance has improved dramatically when important characteristics are extracted using the wrapper method. The LR method identified five significant features with a specificity, sensitivity, accuracy and F1 score of 0.87, 0.33, 87.12% and 0.9309. The region, whether the infant is a twin, and cesarean delivery were the three key features discovered by the DT and RF models, which were implemented using the wrapper technique. All three models had the identical F1 score of 0.9318. However, “child is twin” was recognized as a significant feature by the SVM, NB, and AB models, with an F1 score of 0.9315. Ultimately, with an F1 score of 0.9315, the XGBoost model recognized “child is twin” and “age at first sex” as relevant features. Random Forest again beat the other approaches in this instance.ConclusionsThe study reveals Wrapper method as the optimal feature selection technique. The ML method outperforms traditional methods, with Random Forest (RF) being the most effective predictive model for Low-Birth-Weight prediction. The study suggests that policymakers in Bangladesh can mitigate low birth weight newborns by considering identified risk factors.

The problem persists despite the heavy consequences of jail and the death penalty imposed on those found guilty of selling drugs. This situation creates an alarm among academics and managers who work in practice. Hotbeds for drug dealers are on the rise, and their financial impact on the global economy is moving forward. Drug peddlers distract society by warping peace, justice, and order and threaten the Sustainable Development Goals (SDGs). This study evaluated secondary data containing qualities and suspect drug groups to determine whether machine learning techniques could predict the suspect drug group. We developed a prediction model using nine machine learning algorithms: AdaBoost (AB), naïve Bayes (NB), logistic regression (LR), K nearest neighbour (KNN), random forest (RF), decision tree (DT), neural network (NN), CN2 and support vector machine (SVM). The study utilized Orange data mining software to obtain a more accurate perspective on the data. Predictive accuracy was determined using 5-fold stratified cross-validation. Friedman’s test was conducted, and the results showed that the performance of each algorithm was significantly different. Also, the study compared the models and compiled the results. The results reveal that the random forest has the highest accuracy compared to the others. This prediction model implies that high accuracy can help the government make informed decisions by accurately identifying and classifying suspected individuals and offenders. This prediction will help the government refer suspected individuals and offenders who meet the qualifications for specific drug law sections, like prosecution or rehabilitation, to those sections with a faster and more accurate rate. The outcome of this study should be helpful to law enforcement agents, analysts, and other drug practitioners who may find machine learning tools dependable to detect and classify drug offenders.KeywordsDrug groupClassificationMachine learningOffendersDrug peddling

A machine learning method is proposed to predict the collapse patterns and flow field state of underwater explosion bubbles subjected to the vertical sidewall and free surface, which can overcome the limitations of high costs of traditional experimental tests and long computation times of numerical simulations. The dataset was obtained by the boundary element method, including the cases of the bubble with different buoyancy parameters at different distances from the free surface and vertical sidewall. Due to the strong geometric nonlinearity of the bubble influenced by boundary, three classification models are adopted to identify the collapse patterns of bubbles, which are support vector machines, K nearest neighbor, and decision tree. Meanwhile, an ensemble learning (EL) model based on the three classification models is adopted to enhance the prediction accuracy. Furthermore, three regression models, which are deep neural network (DNN), extreme learning machine (ELM), and random forest (RF), were adopted and compared to predict flow field information around the bubble. The results show that EL exhibits better robustness to the distribution and proportion of samples when identifying collapse patterns. Meanwhile, compared with ELM and RF, DNN demonstrates stronger performance in capturing nonlinear relationships, especially in regions where the bubble curvature changes abruptly. Moreover, a learning rate decay strategy is proposed to effectively suppress the phenomenon of loss oscillation in the training process of DNN based on adaptive activation functions.

For environmental protection, urban planning, monitoring, and management of the urban ecosystem, mapping urban green spaces is a crucial undertaking. A vital source of information for United Nations Sustainable Development Goal 11.7 could come from urban green space mapping. The standard method of mapping urban green spaces requires field measurements and takes a lot of time. It is also important to update urban green space maps periodically because urban green spaces can change quickly over time due to development. With the advent of high-resolution satellite sensors like Sentinel-1 and Sentinel-2, a large number of remote sensing images may be gathered, providing quick and precise information over urban areas. This work intends to offer a new perspective on how crowd sourced geospatial big data and remote sensing may be combined to enhance the mapping of urban green spaces, including time optimization and accurate information through machine learning and deep learning. For the revitalization of cities, this data will be valuable. Remote sensing imagery data can be classified using machine learning techniques like Support vector machines (SVM), Random forests (RF), and Naive Bayes (NB). In deep learning techniques such as Long Short-Term Memory (LSTM), Convolutional Neural Network (CNN), K Nearest Neighbor (KNN), Generative Adversarial Networks (GAN), and Recurrent Neural Networks (RNN) can be used to classify remote sensing images.

The most fatal disease on the earth is thought to be illness of the heart. There are a lot of features that change the heart’s composition or functionality. In most cases, it is hard for doctors to make a diagnosis accurately and quickly. This study’s objective is to determine critical factors and methods of data mining which can improve the accuracy for prediction of heart disease. Further, it is essential to make use of automatic technologies in diagnosing heart diseases as early as possible. To develop a new prediction technique for heart disease that comprises four phases such as “(a) Pre-processing, (b) Feature extraction, (c) Feature selection and (d) Classification”. The initial stage of pre-processing is when the incoming data is treated to the elimination of redundant and missed numbers. Then, from the initial stage of data, the higher-order statistical and statistical characteristics, chi-squared features and symmetrical uncertainty attributes are derived. However, when working with a greater number of characteristics, the curse of dimensionality was a severe issue. Hence, the characteristics of optimal features are planned from the overall set of features. A novel Hybrid Bull and Elephant Algorithm (HB-EA) is introduced for the selection of optimal features. Consequently, the selected set of features is subjected to various classifiers as an ensemble model that contains “Naïve Bayes (NB), Decision Tree (DT), Neural Network (NN), Support Vector Machine (SVM), Optimized Recurrent Neural Network (RNN) and Linear Regression (LR)”. The final step is to log off efficiency for outputs obtained from the group of classifiers and determine the outcome. The RNN weights are ideally tuned by the suggested HB-EA technique to boost the system’s accuracy. The proposed model is finally evaluated against existing techniques to determine its superiority. The suggested technique for dataset 1 achieved maximum accuracy (0.916), and it is 15.24%, 8.76%, 7.56%, 4.09% and 1.89% better than convolution schemes like Random Forest (RF), Deep Belief Network (DBN), SVM, [Formula: see text]-Nearest Neighbor (KNN) and Elephant Herding Optimization (EHO) models.

The nature of stock market movement has always been ambiguous for investors because of various influential factors. This study aims to significantly reduce the risk of trend prediction with machine learning and deep learning algorithms. Four stock market groups, namely diversified financials, petroleum, non-metallic minerals and basic metals from Tehran stock exchange, are chosen for experimental evaluations. This study compares nine machine learning models (Decision Tree, Random Forest, Adaptive Boosting (Adaboost), eXtreme Gradient Boosting (XGBoost), Support Vector Classifier (SVC), Naïve Bayes, K-Nearest Neighbors (KNN), Logistic Regression and Artificial Neural Network (ANN)) and two powerful deep learning methods (Recurrent Neural Network (RNN) and Long short-term memory (LSTM). Ten technical indicators from ten years of historical data are our input values, and two ways are supposed for employing them. Firstly, calculating the indicators by stock trading values as continues data, and secondly converting indicators to binary data before using. Each prediction model is evaluated by three metrics based on the input ways. The evaluation results indicate that for the continues data, RNN and LSTM outperform other prediction models with a considerable difference. Also, results show that in the binary data evaluation, those deep learning methods are the best; however, the difference becomes less because of the noticeable improvement of models’ performance in the second way.

This paper disseminates an extra dimension of substantial analysis demonstrating the trade-offs between the performance of Parametric (P) and Non-Parametric (NP) classification algorithms when applied to classify faults occurring in pneumatic actuators. Owing to the criticality of the actuator failures, classifying faults accurately may lead to robust fault tolerant models. In most cases, when applying machine learning, the choice of existing classifier algorithms for an application is random. This work, addresses the issue and quantitatively supports the selection of appropriate algorithm for non-parametric datasets. For the case study, popular parametric classification algorithms namely: Naïve Bayes (NB), Logistic Regression (LR), Linear Discriminant Analysis (LDA), Perceptron (PER) and non-parametric algorithms namely: Multi-Layer Perceptron (MLP), k Nearest Neighbor (kNN), Support Vector Machine (SVM), Decision Tree (DT) and Random Forest (RF) are implemented over a non-parametric, imbalanced synthetic dataset of a benchmark actuator process. Upon using parametric classifiers, severe adultery in results is witnessed which misleads the interpretation towards the accuracy of the model. Experimentally, about 20% improvement in accuracy is obtained on using non-parametric classifiers over the parametric ones. The robustness of the models is evaluated by inducing label noise varying between 5% to 20%. Triptych analysis is applied to discuss the interpretability of each machine learning model. The trade-offs in choice and performance of algorithms and the evaluating metrics for each estimator are analyzed both quantitatively and qualitatively. For a more cogent reasoning through validation, the results obtained for the synthetic dataset are compared against the industrial dataset of the pneumatic actuator of the sugar refinery, Development and Application of Methods for Actuator Diagnosis in Industrial Control Systems (DAMADICS). The efficiency of non-parametric classifiers for the pneumatic actuator dataset is well proved.

Soil texture identification using LIBS data combined with machine learning algorithm

The healthcare industry struggles with diagnosing health disorders and finding the right treatment for diseases at an earlier stage. Many diseases can be detected early and treated well. Thyroid malfunction is one such disease. One of the most important organs in both men and women is the thyroid gland, a vascular gland. This gland secretes the hormones that help to control the body's metabolism. Hyperthyroidism and hypothyroidism are two types of thyroid ailments. Clinical examination and numerous blood tests are necessary for the traditional methods of detecting thyroid disorders. The main aim is to more precisely identify thyroid dysfunction at an early stage. This research makes a study of a variety of classification approaches, including Decision Tree (DT), Support Vector Machine (SVM), Random Forest (RF), Naive Bayes (NB), Logistic Regression (LR), Linear Discriminant Analysis (LDA), K Nearest Neighbour (KNN) and Multi-Layer Perceptron (MLP), to predict thyroid illness as early as possible. This study majorly focuses on the performance metric accuracy observed from various Machine Learning (ML) classification Algorithms.

Pneumonia is one of the common diseases amongst children in Bangladesh. Many children die from pneumonia in Bangladesh. Pneumonia is an infection that infects the air sacs in one or both lungs. In Bangladesh, nearly 50,000 children die of pneumonia every year. For diseases forecasting, Machine learning algorithms are popular and used extensively. Machine Learning allows us to fulfill such a task with much consequence. We established our dataset from the particular obtainable from our survey. For prognosticating pneumonia, we employed six traditional Machine Learning algorithms. They are K- Nearest Neighbor (KNN), Naive Bayes classifier, Decision Tree, Support Vector Machine (SVM), Neural Network algorithm, and Random Forest. For implementing these algorithms, we applied Scikit-leam, Pandas, NumPy, and for visualizing our data, we have used Matplotlib and seaborn. By proper interpretation, we considered the best performing algorithm for the prediction of pneumonia. We have measured to classify whether pneumonia declines under pneumonia (Positive) and pneumonia (Negative) class. Among all the algorithms, we have chosen the best algorithm which is provided us best accuracy and F1-score. By the best accomplishing algorithm, our model can predict pneumonia quite well.

Machine learning (ML) models are cost-effective tools that can establish a prognosis by considering clinical, genomic, and histological data from breast cancer patients. In this study, we evaluate the performance of 12 ML models in predicting the overall survival of breast cancer patients from “The Cancer Genome Atlas-Breast Cancer” database at 60 months of follow-up. The models evaluated were Logistic Regression (LR), Ridge Classifier (RC), Least Absolute Shrinkage and Selection Operator (LASSO), K-Nearest Neighbors (KNN), Naive Bayes (NB), Linear Discriminant Analysis (LDA), Decision Tree (DT), Multilayer Perceptron (MP), Stochastic Gradient Descent (SGD), Support Vector Machine (SVM), Random Forest (RF), and XGBoost (XGB). The most accurate models were NB (86.76%) and SGD (85.29%). The models with the highest sensitivity were SGD (93.75%), MP (91.67%), and NB (89.89%), while the models with the highest specificity were KNN (95.0%), DT (90.0%), and LASSO (85.0%). The models with the highest area under the ROC curve were LDA (90.83%), LASSO (90.1%), and LR (89.79%). The most relevant variables were “Previous Diagnosis of Cancer,” “Presence of Tumor,” “Ancillary Therapy,” and “Histology.” Our findings demonstrate that AI tools for predicting patient outcomes are a viable option when expensive prognostic genomic tools are not available. According to our results, NB and LASSO show promising performance in predicting the prognosis of breast cancer patients. Further studies with data from developing countries are needed to improve the performance of these tools.

Accurate prediction of agricultural yield is extremely important to ensure food security and cope with the challenges created by climate change and natural disasters. Forecasting agricultural yield is a challenging task due to the complex nature of variables (fertiliser, rainfall, temperature and others) that affect agricultural production. This study employs six supervised machine learning algorithms: Support Vector Machine (SVM), Decision Tree (DT), Random Forest (RF), Multi-Layer Perceptron (MLP), Recurrent Neural Network (RNN), and Convolutional Neural Network (CNN) to build a predictive model using 49 years of historical data (1973-2021) on paddy, wheat, and maize. Model performance was evaluated using Mean Squared Error (MSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), and Root Mean Squared Error (rMSE). Results show that DT and RF models are the most precise with MSE 1% to 5%, MAE 8% to 21%, followed by SVM and CNN. Key predictors of crop yield include area cultivated, capital expenditure, banking expansion, rainfall, temperature, and fertilizers, while irrigation and road network were less significant. The study recommends that farmers prioritize commercial farming, agricultural equipment, and timely availability of fertilizer for application. The Government of Nepal (GoN) should redirect subsidies towards agricultural mechanization, ensure timely supply of fertilizer, and expand banking services in agricultural areas.

This study introduces a machine learning (ML) adaptive interface for classifying transparent conductive oxide (TCO) materials by examining their optical characteristics, such as their wavelength, optical density, absorption, and transmission. Six supervised ML models were considered in this study, namely decision tree, random forest, extreme gradient boosting (XGBoost), support vector classifier (SVC), K -nearest neighbors (KNN), and multilayer perceptron (MLP). These models were evaluated using metrics such as accuracy, precision, recall, F1-score, confusion matrices, receiver operating characteristic (ROC) curves, and learning curves. The comparison findings revealed that XGBoost delivered the best performance, achieving an accuracy of $$96.21\%$$ 96.21 % and a perfect area under the curve (AUC) score of 1.00, effectively distinguishing all five sampled TCO classes: aluminum-doped zinc oxide (AZO), fluorine-doped tin oxide (FTO), indium tin oxide (ITO), magnesium-doped zinc oxide (MZO), and zinc oxide (ZnO). These findings reveal ML’s potential for both scientific and industrial integration in material science research. XGBoost, as a high-precision model, may significantly reduce process time and equipment-intensive laboratory characterizations by allowing for the rapid and precise classification of data from optical measurements. This expedites the search for materials suitable for use in photovoltaics, optoelectronics, and display technologies, all of which require high transparency and conductivity. Overall, this work expands the potential for using ML to categorize materials, particularly XGBoost. It provides both scientific understanding and practical applications in innovative substance informatics and industrial quality control.