A Novel Stacked Ensemble for Hate Speech Recognition

Detection of visual faults in photovoltaic modules using a stacking ensemble approach

Customer satisfaction is one of the factors that can be used to measure the success of service in a company. In the era of the 2000s until now, internet service providers have continued to grow throughout the world, including in Indonesia. IndiHome and First Media are companies that provide internet services that make it easy for the public to communicate and obtain information. With many uses of IndiHome and First Media internet services, there are often several obstacles that cause various responses from users. Users usually channel these responses to IndiHome or First Media customer care on Twitter. The dataset for this study was obtained from Twitter using the Twitter API and the Tweepy library. The dataset that has been collected is 6.962 tweets for the IndiHome dataset and 8,089 tweets for the First Media dataset. This study conducts sentiment analysis using the Ensemble Stacking with three base classifiers and a meta classifier. The base classifier used is Naïve Bayes, K-Nearest Neighbor, and Decision Tree, while the meta classifier used is Logistic Regression. This study uses the term frequency-inverse document frequency (TF-IDF) to determine the frequency value of a word in a document. This study uses two test scenarios: testing without oversampling and testing with oversampling on the dataset. The results show that Ensemble Stacking with term frequency-inverse document frequency feature extraction produces the highest accuracy, with an accuracy value of 88.27% on the IndiHome dataset and 92.56% on the First Media dataset by oversampling on both datasets.

Background/Objectives: Recent studies emphasized on using ensemble models over single ones to solve credit scoring problems. The objective of this study is to build a heterogeneous ensemble classifier model with an improved classification accuracy. Methods: This study focuses on developing a heterogeneous ensemble classifier using Logistic Regression, K-nearest neighbor, Decision tree, Random Forest, Naïve Base and Support vector machine as base classifiers and Random Forest, Logistic Regression and Support vector machine as meta-classifiers. The proposed model is built using these six base classifiers for ensemble aggregation. A feature selection algorithm based on the random forest technique is used for selecting the best features. A stacking and voting method are used for building ensemble model. Findings: The ensemble classifier gives superior predictive performance than single classifiers SVM, DT, RF, NB, KNN and LR with an accuracy of 91.56% for Australian dataset and 84.35% for German dataset. Novelty: The proposed model uses stacking and majority voting method for ensemble classification. Initially, stacking is applied to the base classifiers. This is done in two levels. First the training dataset is split into 10 folds for cross validation. The output of each classifier is taken, and the dataset is updated with the meta-features. In the second level, three meta-classifiers (MC), namely LR, SVM and RF are used. Majority voting is applied to the output of these meta-classifiers for the prediction. Keywords: Credit scoring; ensemble model; SVM; DT; RF; NB; KNN; LR

Genre classification is a process of grouping similarities, such as patterns, styles, or objectives with management data as already in the music (e.g. pop and rock). It is used along with the classification of topics. This paper will classify songs from audio signal to a hierarchy of musical genre by using feature extraction. Trimbral texture, rhythmic content and pitch content are used as the main feature sets. Feature selection is selected by using Particle Swarm Optimization (PSO) and sent selected feature to classification. The result in classification has low accuracy. Thus, using stacking ensemble method is to improve the prediction. In this paper, the purpose is to improve the prediction by using stacking ensemble method. Stacking ensemble that have the second level is base classifier and meta-classifier. In base classifier consists of 5 classification; K-Nearest Neighbors (k-NN), Decision Tree (DT), Random Forest, Support Vector Machines (SVM); and Naive Bayes. This process is generated to build multiple classifier predictors and sent it to meta-classifier. In the process of meta-classifier will create new model to predict test data. The new model has been created from neural network which train data is the output of base classifier.

Predicting stock returns by classifier ensembles

Support Vector Machines (SVMs) ensembles have been widely used to improve classification accuracy in complicated pattern recognition tasks. In this work we propose to apply an ensemble of SVMs coupled with feature-subset selection methods to aleviate the curse of dimensionality associated with expression-based classification of DNA microarray data. We compare the single SVM classifier to SVM ensembles applying two different feature-subset selection techniques, namely random selection and k-means clustering, the base classifiers are combined using either majority vote or SVM fusion. Two real-world benchmarks datasets are used to evaluate and compare the performance. Experimental results show that the SVM ensemble of SVM base classifiers using k-means clustering for feature-subset selection and employing an SVM combiner achieve the best classification accuracy, and that feature-subset-selection methods can have considerable impact on the classification accuracy.

In the cyber world, network security is increasingly challenged by sophisticated attacks and vulnerabilities that exploit data and systems. Effective security solutions are crucial for real-time protection, yet existing methods often struggle with imbalanced datasets, redundant features, and limited classification accuracy. Network Intrusion Detection Systems (NIDS) play a vital role in combating these threats, but traditional ML approaches frequently fall short in detecting complex and subtle attack patterns. This research addresses these critical issues by introducing a Stacked Ensemble-Based Network Intrusion Detection System using Boruta (ASENIDSB). This model tackles the problem of class imbalance, a significant challenge in IDS, by employing the K-means SMOTE oversampling technique to improve the representation of minority classes. Additionally, the Boruta feature selection method improves computational efficiency by addressing redundancy and inefficiency brought on by irrelevant features. ASENIDSB also improves classification robustness and detection capabilities using a stacked ensemble learning technique. It integrates Extra Trees and Extreme Gradient Boosting as base classifiers, with Logistic Regression serving as the meta-model to deliver precise and reliable predictions. This combination enables the model to address the limitations of single and traditional ensemble classifiers, offering a robust, efficient, and accurate solution for network intrusion detection. The accuracy and F1-score of the proposed ASENIDSB model are evaluated using a ten-fold cross-validation method. Experimental Results have proved that the proposed model significantly surpasses some other ensemble models, standard ML classifiers, and SOTA models. The proposed ASENIDSB model achieved an impressive accuracy of 97.85% and an F1-score of 97.82%.

Given the limitations of traditional single prediction models in dealing with complex influencing factors on academic performance in universities, this study proposes an academic performance prediction method based on Stacking multi model fusion. Taking college students majoring in digital media as the research object, a learning engagement measurement scale is introduced to provide targeted data support for model construction, and the principle of Stacking ensemble learning and its advantages in improving model generalization ability and stability are systematically explained. Provide a detailed description of the construction process for Stacking ensemble learning models. Select candidate algorithms from random forest, XGBoost, and KNN machine learning algorithms in the selection of basic learners; Using linear regression algorithm as a meta learner, optimizing model parameters through cross validation strategy; At the same time, strict preprocessing is carried out on the collected data, and raverse all combinations of base classifier algorithms for ensemble training, and construct a Stacking ensemble classifier. In the experimental stage, based on students' grades in digital media courses, the effectiveness of the model is evaluated by comparing the predictive performance of Stacking ensemble classifiers and single classifiers. The experimental results show that the Stacking ensemble classifier performs excellently in prediction accuracy and robustness, significantly better than a single classifier. The R² Score has increased by approximately 10.42% to 16.82%. This model can objectively and accurately reflect students' learning status, providing scientific basis for blended course teaching design, digital media professional teaching management strategy formulation, and personalized guidance for students' academic careers.

With the widespread use of social media, people from all walks of life—individuals, friends, family, public and private organizations, business communities, states, and entire nations—are exchanging information in various formats, including text, messages, audio, video, cartons, and pictures. Social media also facilitates the distribution and propagation of hate speech, despite the immense benefits of knowledge sharing through these platforms. The purpose of this work was to construct a text-based, Pidgin English hate speech classification system (HSCS) in social media, taking into account the alarming rate at which hate speech is shared and propagated on social media, as well as the negative effects of hate speech on society. We used text data sets in Pidgin English that were taken from Twitter and Facebook (3,153). To train the Support Vector Machine (SVM) text classifier to identify hate speech in Pidgin English, 70% of the Pidgin English data set was annotated. The SVM classifier's performance was tested and assessed using the remaining thirty percent of the Pidgin English text data set. The test set findings' confusion matrix, as determined by the HSCS performance evaluation, was 62.04%, 64.42%, 0.7541, 0.6947, and 0.64 in terms of accuracy, precision, recall, F1-score, and Receiver Operating Characteristics (ROC) curve. When HSCS was compared to other Machine Learning (ML) classifiers, such as Logistic Regression (LR), Random Forest (RF), and Naive Bayes, the results showed that LR had accuracy and precision of 61.51% and 63.89%, RF had 54.88% and 50.65%, and Naive Bayes had 61.51% and 63.89%.

This paper introduces a hybrid spatial fusion model for near real-time player recognition in T20i cricket. The system combines ResNet50 for deep feature extraction fused with Support Vector Machine (SVM), K-Nearest Neighbors (KNN) base classifiers, and a Logistic Regression meta-classifier in a stacking ensemble. A novel, domain-specific dataset was developed to capture challenges such as occlusions, variable lighting, and distant camera views followed by an expert evaluation and validation of the dataset. The proposed model achieved 98.14% accuracy, 98% precision, and 98% recall, significantly outperforming standalone deep learning models. These results highlight the advantages of combining deep and traditional machine learning methods. The approach is both robust and resource-efficient, offering a practical solution for sports analytics and laying the foundation for future work on temporal data and transfer learning.

Social media platforms provide an inexpensive communication medium that allows anyone to quickly reach millions of users. Consequently, in these platforms anyone can publish content and anyone interested in the content can obtain it, representing a transformative revolution in our society. However, this same potential of social media systems brings together an important challenge---these systems provide space for discourses that are harmful to certain groups of people. This challenge manifests itself with a number of variations, including bullying, offensive content, and hate speech. Specifically, authorities of many countries today are rapidly recognizing hate speech as a serious problem, specially because it is hard to create barriers on the Internet to prevent the dissemination of hate across countries or minorities. In this paper, we provide the first of a kind systematic large scale measurement and analysis study of hate speech in online social media. We aim to understand the abundance of hate speech in online social media, the most common hate expressions, the effect of anonymity on hate speech and the most hated groups across regions. In order to achieve our objectives, we gather traces from two social media systems: Whisper and Twitter. We then develop and validate a methodology to identify hate speech on both of these systems. Our results identify hate speech forms and unveil a set of important patterns, providing not only a broader understanding of online hate speech, but also offering directions for detection and prevention approaches.

In a text classification problem, imbalance nature in dataset oftentimes has been disregarded even though it might significantly have an impact on the results of the classification model performance. This issue also occurred in the hate speech detection where most collected datasets are highly unbalanced. Among state-of-the-art methods that deal with classifying disparity data, the sampling-based technique is the most effective approach in classifying an imbalanced data. In this paper, four resampling methods include Random Oversampling (ROS), Synthetic Minority Technique (SMOTE), Adaptive Synthetic (ADASYN) and Random Undersampling (RUS) are used as an answer to the inequality of class distribution in a hate speech dataset. With three basic machine learning classifiers i.e. Support Vector Machine, Logistic Regression and Naïve Bayes, the evaluation results show that the oversampling approach improves the accuracy and the overall performance of three classifiers. Among all resampling techniques and machine learning algorithms, Logistic Regression enforced by ROS performed the best with an overall accuracy of 91 percent and F1-Score of 0.95.

With the rising demand for integrated and autonomous systems in the field of engineering, efficient frameworks for instant detection of performance anomalies are imperative for improved productivity and cost-effectiveness. This study proposes a systematic predictive maintenance framework based on the hybrid multisensor fusion technique of fuzzy rough set feature selection and stacked ensemble for the efficient classification of fault conditions characterised by uncertainties. First, a feature vector of time-domain features was extracted from 17 multiple sensor signals. Then, a comparative study of six different Fuzzy Rough Set Feature Selection (FRFS) methods was employed to select the various combinations of optimal feature subsets for various faults classification tasks. The determined optimal feature subsets then served as inputs for training the stacked ensemble (ESB(STK)). In the ESB(STK), Support Vector Machine (SVM), Multilayer Perceptron (MLP), k -Nearest Neighbour ( k -NN), C4.5 Decision Tree (C4.5 DT), Logistic Regression (LR), and Linear Discriminant Analysis (LDA) served as the base classifiers while the LR was selected to be the metaclassifier. The proposed hybrid framework (FRFS-ESB(STK)) improved the classification accuracy with the selected combinations of optimal feature subset size whiles reducing the computational cost, overfitting, training runtime, and uncertainty in modelling. Overall analyses showed that the FRFS-ESB(STK) proved to be generalisable and versatile in the classification of all conditions of four monitored hydraulic components (i.e., cooler, valve, accumulator, and internal pump leakage) when compared with the six base classifiers (standalone) and three existing ensemble classifiers (Stochastic Gradient Boosting (SGB), Ada Boost (ADB), and Bagging (BAG)). The proposed FRFS-ESB(STK) showed an average improvement of 11.28% and 0.88% test accuracies when classifying accumulator and pump conditions, respectively, whiles 100% classification rates were obtained for both cooler and valve.

Social media has become a domain that involves a lot of cyberbullying and hate speech. Some users feel entitled to engage in abusive conversations by sending abusive messages, tweets, or photos to other users. It is critical to detect hate speech and prevent innocent users from becoming victims of cyberbullying. In this paper, we investigate the applicability and performance of a variety of machine learning techniques with the help of text processing techniques in the pursuit of a reliable system for identifying hate speech. We evaluate Naïve Bayes, Support Vector Machines, Decision Trees, Random Forests, Logistic Regression, and K Nearest Neighbors on three different datasets. We used Term Frequency-Inverse Document Frequency (TF-IDF), unigram, bigram, trigram, combination of unigram and bigram, and combination of unigram, bigram, and trigram for the text corpus read by the machine learning techniques to see which approach gives better results. Additionally, since the datasets are unbalanced, we under-sampled and over-sampled the data and investigated the results. Our results show that on all three datasets over-sampling the minority class and using a combination of unigram, bigram, and trigram gave the best results with Random Forests achieving the highest F1-score of at least 0.96 on the three datasets. This was followed by SVM and Logistic Regression, which achieved F1-scores of at least 0.939.

In comparison to other methodologies, Machine Learning has proven to be an effective tool for making reasonable predictions. Initially, a single machine learning model is used on data to make predictions; however, instead of using a single machine learning model, a collection of ML models is employed instead. Models can reduce prediction variance, bias, or both. The result provided will be more appealing and correct if appropriate features affecting the prediction values are supplied. This study presents a framework for selecting feature subsets using the L2 regularization feature selection technique. The study took into account 11 crops and 15 characteristics from 300 districts across India. For predicting yield, basic learners' models such as Linear Regression, LASSO Regression, Decision Trees, Random Forests, Extreme Gradient Boosting (XGBoost), and Support Vector Machines are initially trained and tested. Base learners are used to create the stacking ensemble and the proposed optimum stacking ensemble. In terms of the performance measures employed, the proposed Optimized stacking ensemble shows approximately (MSE =6408, Accuracy=85%) outperforms the Normal stacking ensemble (MSE=8315, Accuracy=83%).