Voting Technique Research Articles

D{S}^2L{C}^3Net : A Decision Support System for Lung Colon Cancer Classification using Fusion of Deep Neural Networks and Normal Distribution based Gray Wolf Optimization

Lung and Colorectal (LC) cancer is life-threatening and rapidly developing cancers. According to World Health Organization (WHO), approximately 4.14 million lung and colorectal cancer cases were newly diagnosed, with 2.7 million fatalities. An International Agency for Research on Cancer (IARC) reported that there will be more than 3 million additional instances of colorectal cancer worldwide between 2020 and 2040. Early diagnosis of LC cancer is very helpful for treatment and can save the precious human life. The conventional diagnosis methods are expensive and time consuming. In this work, we present an accurate and efficient model for the classification of Lung and Colorectal (LC) cancer. We utilize two well-known pre-trained deep learning models, ResNet50 and EfficientNetB0, and fine-tuned the both models based on the addition and removal of layers. After the fine-tuning, manual hit and trail based hyperparameters are initialized. Later on, the deep transfer learning was performed and obtained the trained models. Two different feature vectors have been extracted from both models and fused using a priority based serial approach. To further improve the performance of extracted features, Normal Distribution based Gray Wolf Optimization algorithm is employed and obtained the best features that given as input to five classifiers. The output of these five classifiers is then utilized by soft voting technique to generate the final prediction. Experimental results show that the proposed architecture achieved an overall 98.73% accuracy on LC25000 dataset. Furthermore, prediction time was reduced by 19.14%. Comparison with the state-of-the-art techniques shows that the proposed technique obtained the improved performance results.

Predicting patients with Parkinson's disease using Machine Learning and ensemble voting technique

Predicting patients with Parkinson's disease using Machine Learning and ensemble voting technique

Arab reactions towards Russo-Ukrainian war

The aim of this paper is to analyze the Arab peoples reactions and attitudes towards the Russo-Ukraine War through the social media of posted tweets, as a fast means to express opinions. We scrapped over 3 million tweets using some keywords that are related to the war and performed sentiment, emotion, and partiality analyses. For sentiment analysis, we employed a voting technique of several pre-trained Arabic language foundational models. For emotion analysis, we utilized a pre-constructed emotion lexicon. The partiality is analyzed through classifying tweets as being ‘Pro-Russia’, ‘Pro-Ukraine’, or ‘Neither’; and it indicates the bias or empathy towards either of the conflicting parties. This was achieved by constructing a weighted lexicon of n-grams related to either side. We found that the majority of the tweets carried ‘Negative’ sentiment. Emotions were not that obvious with a lot of tweets carrying ‘Mixed Feelings’. The more decisive tweets conveyed either ‘Joy’ or ‘Anger’ emotions. This may be attributed to celebrating victory (‘Joy’) or complaining from destruction (‘Anger’). Finally, for partiality analysis, the amount of tweets classified as being ‘Pro-Ukraine’ was slightly greater than Pro-Russia’ at the beginning of the war (specifically from Feb 2022 till April 2022) then slowly began to decrease until they nearly converged at the start of June 2022 with a shift happening in the empathy towards Russia in August 2022. Our Interpretation for that is with the initial Russian fierce and surprise attack at the beginning and the amount of refugees who escaped to neighboring countries, Ukraine gained much empathy. However, by April 2022, Russian intensity has been decreased and with heavy sanctions the U.S. and West have applied on Russia, Russia has begun to gain such empathy with decrease on the Ukrainian side.

Ensemble Wavelet Decomposition-Based Detection of Mental States Using Electroencephalography Signals.

Technological advancements in healthcare, production, automobile, and aviation industries have shifted working styles from manual to automatic. This automation requires smart, intellectual, and safe machinery to develop an accurate and efficient brain-computer interface (BCI) system. However, developing such BCI systems requires effective processing and analysis of human physiology. Electroencephalography (EEG) is one such technique that provides a low-cost, portable, non-invasive, and safe solution for BCI systems. However, the non-stationary and nonlinear nature of EEG signals makes it difficult for experts to perform accurate subjective analyses. Hence, there is an urgent need for the development of automatic mental state detection. This paper presents the classification of three mental states using an ensemble of the tunable Q wavelet transform, the multilevel discrete wavelet transform, and the flexible analytic wavelet transform. Various features are extracted from the subbands of EEG signals during focused, unfocused, and drowsy states. Separate and fused features from ensemble decomposition are classified using an optimized ensemble classifier. Our analysis shows that the fusion of features results in a dimensionality reduction. The proposed model obtained the highest accuracies of 92.45% and 97.8% with ten-fold cross-validation and the iterative majority voting technique. The proposed method is suitable for real-time mental state detection to improve BCI systems.

A Decision-Fusion-Based Ensemble Approach for Malicious Websites Detection

Malicious websites detection is one of the cyber-security tasks that protects sensitive information such as credit card details and login credentials from attackers. Machine learning (ML)-based methods have been commonly used in several applications of cyber-security research. Although there are some methods and approaches proposed in the state-of-the-art studies, the advancement of the most effective solution is still of research interest and needs to be improved. Recently, decision fusion methods play an important role in improving the accuracy of ML methods. They are broadly classified based on the type of fusion into a voting decision fusion technique and a divide and conquer decision fusion technique. In this paper, a decision fusion ensemble learning (DFEL) model is proposed based on voting technique for detecting malicious websites. It combines the predictions of three effective ensemble classifiers, namely, gradient boosting (GB) classifier, extreme gradient boosting (XGB) classifier, and random forest (RF) classifier. We use these classifiers because their advantages to perform well for class imbalanced and data with statistical noises such as in the case of malicious websites detection. A weighted majority-voting rule is utilized for generating the final decisions of used classifiers. The experimental results are conducted on a publicly available large dataset of malicious and benign websites. The comparative study exposed that the DFEL model achieves high accuracies, which are 97.25% on average of 10-fold cross-validation test and 98.50% on a holdout of 30% test set. This confirms the ability of proposed approach to improve the detection rate of malicious websites.

Hybrid ensemble-based machine learning model for predicting phosphorus concentrations in hydroponic solution

Accurate, label-free, and rapid methods for measuring phosphorus concentrations are essential in a hydroponic system, as excessive or insufficient phosphorus levels can adversely affect plant growth, human health, and environmental sustainability. In this study, we demonstrate the advantages of hybrid machine learning models compared to single machine learning models in predicting phosphorus concentration based on the absorbance dataset. Three machine learning classifiers- Random Forest (RF), Support Vector Machine (SVM), and K-Nearest Neighbors (KNN)- were employed as bases for single and hybrid machine learning models. Three ensemble techniques (voting, bagging, and stacking) were used to hybridize the classifiers. Among the single models, KNN demonstrated the fastest computational time of 18.07 s, while SVM achieved the highest accuracy of 99.6%. The hybrid SVM/KNN model using a voting classifier showed a significant increase in accuracy for KNN with only a slight increase in computational time. Bagging techniques increased the accuracy but at a longer computational time. The stacking technique, which combined SVM, KNN, and RF, achieved the highest accuracy of 99.73% with a short computational time of 36.18 s compared to the bagging and voting technique. This study demonstrates that the machine learning method can effectively distinguish phosphorus concentrations. In contrast, hybrid machine learning techniques can improve accuracy for predicting phosphorus without using labels, despite requiring longer computational time.

Symmetry Breaking for Voting Mechanisms.

Recently, Rowe and Aishwaryaprajna (2019) introduced a simple majority vote technique that efficiently solves Jump with large gaps, OneMax with large noise, and any monotone function with a polynomial-size image. In this paper, we identify a pathological condition for this algorithm: the presence of spin-flip symmetry in the problem instance. Spin-flip symmetry is the invariance of a pseudo-Boolean function to complementation. Many important combinatorial optimization problems admit objective functions that exhibit this pathology, such as graph problems, Ising models, and variants of propositional satisfiability. We prove that no population size exists that allows the majority vote technique to solve spin-flip symmetric functions of unitation with reasonable probability. To remedy this, we introduce a symmetry-breaking technique that allows the majority vote algorithm to overcome this issue for many landscapes. This technique requires only a minor modification to the original majority vote algorithm to force it to sample strings in {0,1}n from a dimension n-1 hyperplane. We prove a sufficient condition for a spin-flip symmetric function to possess in order for the symmetry-breaking voting algorithm to succeed, and prove its efficiency on generalized TwoMax, a spin-flip symmetric variant of Jump, and families of constructed 3-NAE-SAT and 2-XOR-SAT formulas. We also prove that the algorithm fails on the one-dimensional Ising model, and suggest different techniques for overcoming this. Finally, we present empirical results that explore the tightness of the runtime bounds and the performance of the technique on randomized satisfiability variants.

Deteriorated image classification model for malayalam palm leaf manuscripts

The method for document image classification presented in this paper mainly focuses on six different Malayalam palm leaf manuscripts categories. The proposed approach consists of three phases: dataset analysis, building a bag of words repository followed by recognition and classification using a voting approach. The palm leaf manuscripts are initially subject to pre-processing and subjective analysis techniques to create a bag of words repository during the dataset analysis phase. Next, the textual components from the manuscripts are extracted for recognition using Tesseract 4 OCR with default and self-adapted training sets and a deep-learning algorithm. The Bag of Words approach is used in the third phase to categorize the palm leaf manuscripts based on textual components recognized by OCR using a voting process. Experimental analysis was done to analyze the proposed approach with and without the voting techniques, varying the size of the Bag of Words with default/self-adapted training datasets using Tesseract OCR and a deep learning model. Experimental analysis proves that the proposed approach works equally well with/ without voting with a bag of words technique using Tesseract OCR. It is noticed that, for document classification, an overall accuracy of 83% without voting and 84.5% with voting is achieved with an F-score of 0.90 in both cases using Teserract OCR. Overall, the proposed approach proves to be high generalizable based on trial wise experiments with Bag of Words, offering a reliable way for classifying deteriorated Malayalam handwritten palm manuscripts.

Comparative performance of ensemble machine learning for Arabic cyberbullying and offensive language detection

Since cyberbullying impacts both individual victims and entire society, research on abusive language and its detection has attracted attention in recent years. Because social media sites like Facebook, Instagram, Twitter, and others are so widely accessible, hate speech, bullying, sexism, racism, aggressive material, harassment, poisonous comments, and other types of abuse have all substantially increased. Due to the critical requirement to detect, regulate, and limit the spread of harmful content on social networking sites, we conducted this study to automate the detection of offensive language or cyberbullying. We created a new Arabic balanced data set to be used in the offensive detection process because having a balanced data set for a model would result in improved accuracy models. Recently, the performance of single classifiers has been improved using ensemble machine learning. The purpose of this study is to examine the effectiveness of several single and ensemble machine learning algorithms in identifying Arabic text that contains foul language and cyberbullying. Applying them to three Arabic datasets, we have selected three machine learning classifiers and three ensemble models for this aim. Two of them are offensive datasets that are readily accessible in the public, while the third one was created. The results showed that the single learner machine learning strategy is inferior to the ensemble machine learning methodology. Voting performs is the best performing trained ensemble machine learning classifier, outperforming the best single learner classifier (65.1%, 76.2%, and 98%) for the same datasets with accuracy scores of (71.1%, 76.7%, and 98.5%) for each of the three datasets used. Finally, we improve the voting technique’s performance through hyperparameter tuning on the Arabic cyberbullying data set.

Encrypted network traffic classification based on machine learning

Encrypted traffic is an essential part of maintaining the security and privacy of data transmission. It plays an important role in keeping our networks secure by preventing attackers from intercepting confidential information, which they may access without authorization; However, its effectiveness relies heavily on accurate classification techniques being applied correctly, so we can differentiate between legitimate users' activities versus those attempting malicious activity within the networks’ boundaries. Encrypted network traffic is becoming increasingly common in modern communication systems, presenting a challenge for effective network management and security. To address this challenge, machine learning models have been employed to classify encrypted traffic but with limited success due to the lack of clear visibility into packet contents and an inability to inspect their content. For the sake of tackling this issue, more effective research has begun on developing machine learning models for classifying encrypted payloads without relying on inspecting their contents directly. This research will investigate how features like packet length, time stamps or transport layer security (TLS) and encrypted payload information can be used as input features when attempting classification tasks, instead of analyzing unencrypted content directly from packets themselves which would otherwise be impossible given the current technology constraints. The evaluation process will focus on assessing different model architectures, as well as feature selection techniques that yield improved results over the existing approaches. In this paper, we proposed three approaches to identify encrypted traffic and classify different applications such as browsing, VOIP, file transfer and video streaming. The first two techniques consist of two stages: the first stage is either a neural network or a bi-directional LSTM, and the second stage is a selection of different classification techniques, namely Random Forest, Support vector machine, Linear regression, and K-nearest neighbor. The final result is achieved using an ensemble voting technique. As for the third technique, the network packets are grouped together by Source IP, destination IP and session time before feeding them into three different combinations of LSTM networks; either coupled with convolution 1D or 2D layers, or without. Like the first two techniques, the final result is achieved by means of ensemble voting. Through extensive comparison between the three approaches, The first approach yielded the highest accuracy. However, the performance of the second and third techniques in terms of time complexity was superior. The achieved accuracies were 96.8%, 95.2% and 96.5% for the proposed techniques, respectively.

Restrictive Voting Technique for Faces Spoofing Attack

Face anti-spoofing has become widely used due to the increasing use of biometric authentication systems that rely on facial recognition. It is a critical issue in biometric authentication systems that aim to prevent unauthorized access. In this paper, we propose a modified version of majority voting that ensembles the votes of six classifiers for multiple video chunks to improve the accuracy of face anti-spoofing. Our approach involves sampling sub-videos of 2 seconds each with a one-second overlap and classifying each sub-video using multiple classifiers. We then ensemble the classifications for each sub-video across all classifiers to decide the complete video classification. We focus on the False Acceptance Rate (FAR) metric to highlight the importance of preventing unauthorized access. We evaluated our method using the Replay Attack dataset and achieved a zero FAR. We also reported the Half Total Error Rate (HTER) and Equal Error Rate (EER) and gained a better result than most state-of-the-art methods. Our experimental results show that our proposed method significantly reduces the FAR, which is crucial for real-world face anti-spoofing applications.

A novel hybrid ensemble learning for anomaly detection in industrial sensor networks and SCADA systems for smart city infrastructures

Critical Infrastructures (CIs) use Supervisory Control and Data Acquisition (SCADA) systems for monitoring and remote control. Sensor networks are being integrated into all areas of the infrastructures of smart cities. The sensor network data stream contains information that can be utilized to model and control the activity of these infrastructures. However, SCADA systems are constantly exposed to a variety of diverse intrusions, making detection with traditional intrusion detection systems (IDS) extremely difficult. Due to their unique specifications, conventional security solutions, like antivirus and firewall software, are unsuitable for properly securing SCADA systems. In addition, anomaly detection in industrial sensor networks (ISNs) should occur in real time. Therefore, effectively identifying cyberattacks in major SCADA systems is unquestionably essential for enhancing their resilience, ensuring safe operations, and avoiding expensive maintenance. We developed a novel hybrid ensemble model approach to address these issues. This paper's primary objective is to detect hostile intrusions that have already circumvented firewalls and typical IDS. In this paper, we propose a hybrid Ensemble Learning Model (ELM) for intrusion detection in SCADA systems with ISNs utilizing a tangible data gathered from a gas pipeline system given by Mississippi State University (MSU), the water system, and the high-dimensional University of New South Wales-NB 2015 (UNSW-NB15) data that reflects a typical attack in the Internet of Things (IoT) environment. The unity normalization method was adopted for data preprocessing, and the Principal Component Analysis (PCA) was utilized for feature extraction of the high-dimensional datasets. Grey Wolf Optimizer (GWO) was used for optimizing the bagging, stacking, Adaboost, and an ensemble of classifiers Naive Bayes and Support Vector Machine with a majority voting technique. Then, we utilized the proposed approach founded on the bijective soft-set approach for efficient ELM selection. The experiment was conducted in two phases: Initially, without PCA + GWO for feature extraction and selection on the ELM, and subsequently, with PCA + GWO for feature extraction and selection on the ELM. PCA + GWO on the ensemble of classifiers NB + SVM provided an accuracy of 99%, precision of 100%, recall of 100%, and detection rate of 99.90%, outpacing the ensemble of classifiers without PCA feature extraction and GWO optimization approaches.

TILLAGE : An Ensemble-Based Agricultural Crop Recommendation System

India is primarily an agricultural nation, and both the Indian economy and people's daily lives are heavily reliant on agriculture. In our research One flaw we discovered was that many of them thinks on one aspect (weather or soil) to judge whether crops would grow successfully.This paper presents the precision agricultural recommendation system with more parameters helps to farmer to choose right crops for their field with the help of ensemble majority voting techniques like SVM, Naive bayes, Random Forest, KNN and a Convolutional neural network and also it provide valid fertilizers and pesticide to reduce the farmer’s burden in selection right things to their field.

Customer Analysis Using Machine Learning-Based Classification Algorithms for Effective Segmentation Using Recency, Frequency, Monetary, and Time

Customer segmentation has been a hot topic for decades, and the competition among businesses makes it more challenging. The recently introduced Recency, Frequency, Monetary, and Time (RFMT) model used an agglomerative algorithm for segmentation and a dendrogram for clustering, which solved the problem. However, there is still room for a single algorithm to analyze the data's characteristics. The proposed novel approach model RFMT analyzed Pakistan's largest e-commerce dataset by introducing k-means, Gaussian, and Density-Based Spatial Clustering of Applications with Noise (DBSCAN) beside agglomerative algorithms for segmentation. The cluster is determined through different cluster factor analysis methods, i.e., elbow, dendrogram, silhouette, Calinsky-Harabasz, Davies-Bouldin, and Dunn index. They finally elected a stable and distinctive cluster using the state-of-the-art majority voting (mode version) technique, which resulted in three different clusters. Besides all the segmentation, i.e., product categories, year-wise, fiscal year-wise, and month-wise, the approach also includes the transaction status and seasons-wise segmentation. This segmentation will help the retailer improve customer relationships, implement good strategies, and improve targeted marketing.

Analysis of Birth Data using Ensemble Modeling Techniques

ABSTRACT Machine learning and data mining are being used in different fields like data analysis, prediction, image processing, etc., and particularly in healthcare. Over the past decade, several types of research have been carried out focusing on machine learning and data mining application to generate intuitions from historical data and make predictions about the results. Machine learning algorithms play a vital role in improving healthcare systems due to continuous research in machine learning applications. Several researchers have used algorithms of machine learning to develop systems for decision support, analyze clinical aspects, use historical data to extract useful information, make future predictions and categorize diseases, etc. to help physicians make better decisions. In this study, we used an ensemble modeling voting technique for the classification of the birth dataset. Ensemble models combine individual machine learning algorithms to improve the accuracy by predicting from the combined output of the base classifiers. Gradient boosting classifier (GBC), random forest (RF), bagging classifier (BC), and extra trees classifier (ETC) were used as base learners for making a voting ensemble model for the classification of the birth dataset. The results produced have shown that the voting classifier of support vector machine (SVM), random forest (RF), extra trees classifier, and bagging classifier has given the best results with the proportion of 94.78%, gradient boosting classifier has 84.39% accuracy, the random forest has 94.26% accuracy, extra trees classifier have 94.02% accuracy and bagging classifier has 93.65% accuracy. The accuracy achieved by ensemble modeling is far higher than the machine learning algorithms. Ensemble models increase the accuracy of machine learning algorithms by reducing variance and classification errors. The development of such a system will not only help health organizations to take effective measures to improve the maternal health assessment process but will also open the doors for interdisciplinary research in two different fields in the region.

Patterns of structure-function association in normal aging and in Alzheimer's disease: Screening for mild cognitive impairment and dementia with ML regression and classification models.

The combined analysis of imaging and functional modalities is supposed to improve diagnostics of neurodegenerative diseases with advanced data science techniques. To get an insight into normal and accelerated brain aging by developing the machine learning models that predict individual performance in neuropsychological and cognitive tests from brain MRI. With these models we endeavor to look for patterns of brain structure-function association (SFA) indicative of mild cognitive impairment (MCI) and Alzheimer's dementia. We explored the age-related variability of cognitive and neuropsychological test scores in normal and accelerated aging and constructed regression models predicting functional performance in cognitive tests from brain radiomics data. The models were trained on the three study cohorts from ADNI dataset-cognitively normal individuals, patients with MCI or dementia-separately. We also looked for significant correlations between cortical parcellation volumes and test scores in the cohorts to investigate neuroanatomical differences in relation to cognitive status. Finally, we worked out an approach for the classification of the examinees according to the pattern of structure-function associations into the cohorts of the cognitively normal elderly and patients with MCI or dementia. In the healthy population, the global cognitive functioning slightly changes with age. It also remains stable across the disease course in the majority of cases. In healthy adults and patients with MCI or dementia, the trendlines of performance in digit symbol substitution test and trail making test converge at the approximated point of 100 years of age. According to the SFA pattern, we distinguish three cohorts: the cognitively normal elderly, patients with MCI, and dementia. The highest accuracy is achieved with the model trained to predict the mini-mental state examination score from voxel-based morphometry data. The application of the majority voting technique to models predicting results in cognitive tests improved the classification performance up to 91.95% true positive rate for healthy participants, 86.21%-for MCI and 80.18%-for dementia cases. The machine learning model, when trained on the cases of this of that group, describes a disease-specific SFA pattern. The pattern serves as a "stamp" of the disease reflected by the model.

Large dataset partitioning using ensemble partition-based clustering with majority voting technique

<span lang="EN-US">Large datasets have become useful in data mining for processing, storing, and handling vast amounts of data. However, handling and processing large datasets is time-consuming and memory intensive. As a result, the researchers adopted a partitioning strategy to improve controllability and performance and reduce the time and memory required to handle large datasets. Unfortunately, the numerous clustering techniques available in the literature could confuse experts in choosing the best techniques for a given dataset. Furthermore, no clustering technique can tackle all problems, such as cluster structure, noise, or density. To manage large datasets, existing clustering techniques need scalable solutions. Therefore, this paper proposes an ensemble partition-based clustering with a majority voting technique for large dataset partitioning using the aggregation of k-means, k-medoids, fuzzy c-means, expectation-maximization (EM) and density-based spatial clustering of applications with noise (DBSCAN) techniques. These techniques cluster the large dataset individually in the first stage. The final clusters are discovered in the next stage through a majority voting technique among the five clustering algorithms. These five clustering algorithms assigned data instances to the cluster with the most votes. The experimental findings demonstrate that the ensemble partition-based clustering method surpasses the other five clustering algorithms in terms of execution time and accuracy.</span>

A Meta-Model to Predict and Detect Malicious Activities in 6G-Structured Wireless Communication Networks

The rapid leap in wireless communication systems incorporated a plethora of new features and challenges that accompany the era of 6G and beyond being investigated and developed. Recently, machine learning techniques were widely deployed in many fields, especially wireless communications. It was used to improve network traffic performance regarding resource management, frequency spectrum optimization, latency, and security. The studies of modern wireless communications and anticipated features of ultra-densified ubiquitous wireless networks exposed a risky vulnerability and showed a necessity for developing a trustworthy intrusion detection system (IDS) with certain efficiency/standards that have not yet been achieved by current systems. IDSs lack acceptable immunity against repetitive, updatable, and intelligent attacks on wireless communication networks, significantly concerning the modern infrastructure of 6G communications, resulting in low accuracies/detection rates and high false-alarm/false-negative rates. For this objective principle, IDS system complexity was reduced by applying a unique meta-machine learning model for anomaly detection networks was developed in this paper. The five main stages of the proposed meta-model are as follows: the accumulated datasets (NSL KDD, UNSW NB15, CIC IDS17, and SCE CIC IDS18) comprise the initial stage. The second stage is preprocessing and feature selection, where preprocessing involves replacing missing values and eliminating duplicate values, leading to dimensionality minimization. The best-affected subset feature from datasets is selected using feature selection (i.e., Chi-Square). The third step is represented by the meta-model. In the training dataset, many classifiers are utilized (i.e., random forest, AdaBoosting, GradientBoost, XGBoost, CATBoost, and LightGBM). All the classifiers undergo the meta-model classifier (i.e., decision tree as the voting technique classifier) to select the best-predicted result. Finally, the classification and evaluation stage involves the experimental results of testing the meta-model on different datasets using binary-class and multi-class forms for classification. The results proved the proposed work’s high efficiency and outperformance compared to existing IDSs.

A Hybrid Generic Framework for Heart Problem Diagnosis Based on a Machine Learning Paradigm.

The early, valid prediction of heart problems would minimize life threats and save lives, while lack of prediction and false diagnosis can be fatal. Addressing a single dataset alone to build a machine learning model for the identification of heart problems is not practical because each country and hospital has its own data schema, structure, and quality. On this basis, a generic framework has been built for heart problem diagnosis. This framework is a hybrid framework that employs multiple machine learning and deep learning techniques and votes for the best outcome based on a novel voting technique with the intention to remove bias from the model. The framework contains two consequent layers. The first layer contains simultaneous machine learning models running over a given dataset. The second layer consolidates the outputs of the first layer and classifies them as a second classification layer based on novel voting techniques. Prior to the classification process, the framework selects the top features using a proposed feature selection framework. It starts by filtering the columns using multiple feature selection methods and considers the top common features selected. Results from the proposed framework, with 95.6% accuracy, show its superiority over the single machine learning model, classical stacking technique, and traditional voting technique. The main contribution of this work is to demonstrate how the prediction probabilities of multiple models can be exploited for the purpose of creating another layer for final output; this step neutralizes any model bias. Another experimental contribution is proving the complete pipeline's ability to be retrained and used for other datasets collected using different measurements and with different distributions.

Email-Based Cyberstalking Detection On Textual Data Using Multi-Model Soft Voting Technique Of Machine Learning Approach

ABSTRACT In the virtual world, many internet applications are used by a mass of people for several purposes. Internet applications are the basic needs of people in the modern days of lifestyle which are also making habitual society. Like social media, e-mail technology is also more prevalent among people of different categories for personal and official communications. The widespread use of e-mail-based communication is also raising various types of cybercrimes, including cyberstalking. Cyberstalkers also use an e-mail-based approach to harass the victim in the form of cyberstalking. Cyberstalkers utilize several content-wise and intent-wise approaches to target the victim, such as spamming, phishing, spoofing, malicious, defamatory, e-mail bombing, and non-spam e-mails, including sexism, racism, and threatening, and finally, trying to hack the account over e-mail technology. This paper proposed an EBCD model for automatic cyberstalking detection on textual data of e-mail using the multi-model soft voting technique of the machine learning approach. Initially, experimental works were performed to train, test, and validate all classifiers of three model sets on three different labeled datasets. Dataset D1 contains spam, fraudulent, and phishing e-mail subject, dataset D2 contains spam e-mail body text, while dataset D3 contains harassment-related data. After that, trained, tested, and validated classifiers of all model sets were applied as a combined approach to automatically classify the unlabeled e-mails from the user’s mailbox using the multi-model soft voting technique. The proposed EBCD model successfully classifies the e-mails from the user’s mailbox into cyberstalking e-mails, suspicious e-mails (spam and fraudulent), and normal e-mails. In each model set of the EBCD model, several classifiers, namely support vector machine, random forest, naïve bayes, logistic regression, and soft voting, were used. The final decision in classifying the e-mails from the user’s mailbox was taken by the soft voting technique of each model set. The TF-IDF feature extraction method was used with the entire applied machine learning model sets to obtain the feature vectors from the data. Experimental results show that the soft voting technique not only enhances the performance of the e-mail classification task but also supports making the right decision to avoid the wrong classification. Overall performance of the soft voting technique was better than other classifiers, although the performance of the support vector machine was also notable. As per experimental results, the soft voting technique obtained an accuracy of 97.7%, 97.7%, 98.9%, a precision of 97%, 98.3%, 98.6%, recall of 98.3%, 96.5%, 99.1%, f-score of 97.6%, 97.4%, 98.9%, and AUC of 99.4%, 99.7%, 99.9% on dataset D1, D2, and D3 respectively. The average performance of soft voting of each model set on classified e-mails from the user’s mailbox was also notable, with an accuracy of 96.3%, precision of 98.1%, recall of 94%, f-score of 95.9%, and AUC of 96.8%.