ML Models Research Articles

Brain tumor classification utilizing pixel distribution and spatial dependencies higher-order statistical measurements through explainable ML models

Brain tumors are among the most fatal and devastating diseases, and they often result in a significant reduction in life expectancy. The devising of treatment plans that can extend the lives of affected individuals hinges on an accurate diagnosis of these tumors. Identifying and analyzing large volumes of magnetic resonance imaging (MRI) data manually proves to be both challenging and time-consuming. As a result, there exists a pressing need for a reliable machine-learning approach to accurately diagnose brain tumors, and numerous methods have already been proposed over the last decade. In this paper, a novel, comprehensive approach is proposed for identifying and classifying a given MR brain image as abnormal. Three common brain diseases, namely glioma, meningioma, and pituitary tumor, are chosen as abnormal brains, and the Figshare MRI brain image dataset was collected from the Kaggle and IEEE websites. The proposed method is initiated by employing 1st-order statistics, 2nd-order statistics, and higher-order transformed (DWT) feature extraction to extract features from images. Then missing data is addressed and handled using KNNImputer, followed by the application of the ExtratreesClassifier and PCA feature selection methods to identify the most relevant features and reduce the dimensions of these features. Subsequently, the reduced features are submitted to seven machine learning models, namely RF, GB, CB, SVM, LGBM, DT, and LR. The strategy of k-fold cross-validation is utilized to enhance the performance of those models. Finally, the models are evaluated using XAI approaches, which ensure transparent decision-making processes and provide insights into the model’s predictions. Remarkably, our approach achieves the highest accuracy, precision, recall, F1 score, MCC, Kappa, AUC-ROC, and R2, as well as the lowest loss, among the seven models evaluated, proving its effectiveness and applicability in multiple analytic applications relying on publicly available datasets.

Human-machine interaction with intelligent machine learning algorithms for smart classrooms

ABSTRACT The modern technology of learning and teaching in education has moved on to smart classes. The introduction of artificial intelligence (AI) in the education sector has essentially taken over traditional classrooms and transformed how education leads to admiration. Various data types, such as videos, images, sound, text, and biometric data like eye-tracking potentially captured through smart classrooms, pose new issues and challenges for artificial intelligence algorithms. Hence, this paper suggests the Machine Learning Assisted Intelligent Human-Machine Interaction (MLIHMI) model for the smart classroom monitoring system. The real-time interactive dimension signifies the capability to support the Smart classroom’s teaching and human-computer interactions, including smooth interaction, convenient operation, and interactive monitoring. This paper used ML and HMI models to introduce a new effective database for online learning and smart classroom environments utilising the learners’ body postures, facial expressions, and hand gestures. The numerical findings show that the suggested system achieves less error rate of 10.1% and enhances the accuracy rate of 96.5%, with the highest classification ratio of 95.6%, prediction rate of 93.6%, the detection rate of 97.3%, and engagement rate of 81.3% in student gesture and postures recognition and the identification of facial expressions with the high performance of 93.3% when compared to other existing models.

Securing constrained IoT systems: A lightweight machine learning approach for anomaly detection and prevention

With the advent of advanced technological developments such as IoT, edge, and fog computing, cyber attacks have become increasingly sophisticated. IoT networks facilitate collaborative and intelligent tasks across various domains, including Industry 4.0, digital healthcare, and home automation. However, the proliferation of IoT devices has raised concerns about severe attacks, particularly those targeting resource constraints such as energy and memory. In response to these challenges, Tiny Machine Learning (TinyML) has emerged as a new research area, focusing on machine learning techniques tailored for embedded and IoT systems. This study proposes an ML detection mechanism designed to categorize and detect resource-constrained attacks in IoT devices. We consider IoT devices to be integral components within the continuum of edge and cloud computing, leveraging EdgeML and CloudML for detection purposes. Our paper conducts a comparative analysis of ML models, with a specific focus on energy consumption and memory usage in IoT applications. We compare various ML methodologies, including cloud-based, edge-based, and device-based strategies for both training and detection. The evaluation encompasses the application of these ML techniques to petite IoT devices, utilizing TinyML, as well as cloud and edge devices. Our findings reveal that the Decision Tree algorithm deployed on smart devices surpasses other approaches in terms of training efficiency, resource utilization, and the ability to detect resource-constrained attacks on IoT devices. We demonstrate a high level of accuracy, exceeding 96.9%, across all presented ML models in detecting resource constraint attacks within IoT systems. In summary, this research serves as a guide for implementing effective security measures in the dynamic landscape of IoT and associated technologies.

AI-Driven Insight into Polycarbonate Synthesis from CO2: Database Construction and Beyond

Recent advancements in materials science have garnered significant attention within the research community. Over the past decade, substantial efforts have been directed towards the exploration of innovative methodologies for developing new materials. These efforts encompass enhancements to existing products or processes and the design of novel materials. Of particular significance is the synthesis of specific polymers through the copolymerization of epoxides with CO2. However, several uncertainties emerge in this chemical process, including challenges associated with successful polymerization and the properties of the resulting materials. These uncertainties render the design of new polymers a trial-and-error endeavor, often resulting in failed outcomes that entail significant financial, human resource, and time investments due to unsuccessful experimentation. Artificial Intelligence (AI) emerges as a promising technology to mitigate these drawbacks during the experimental phase. Nonetheless, the availability of high-quality data remains crucial, posing particular challenges in the context of polymeric materials, mainly because of the stochastic nature of polymers, which impedes their homogeneous representation, and the variation in their properties based on their processing. In this study, the first dataset linking the structure of the epoxy comonomer, the catalyst employed, and the experimental conditions of polymerization to the reaction’s success is described. A novel analytical pipeline based on ML to effectively exploit the constructed database is introduced. The initial results underscore the importance of addressing the dimensionality problem. The outcomes derived from the proposed analytical pipeline, which infer the molecular weight, polydispersity index, and conversion rate, demonstrate promising adjustment values for all target parameters. The best results are measured in terms of the (Determination Coefficient) R2 between real and predicted values for all three target magnitudes. The best proposed solution provides a R2 equal to 0.79, 0.86, and 0.93 for the molecular weight, polydispersity index, and conversion rate, respectively. The proposed analytical pipeline is automatized (including AutoML techniques for ML models hyperparameter tuning), allowing easy scalability as the database grows, laying the foundation for future research.

EEG-based optimization of eye state classification using modified-BER metaheuristic algorithm

This article introduces the Modified Al-Biruni Earth Radius (MBER) algorithm, which seeks to improve the precision of categorizing eye states as either open (0) or closed (1). The evaluation of the proposed algorithm was assessed using an available EEG dataset that applied preprocessing techniques, including scaling, normalization, and elimination of null values. The MBER algorithm’s binary format is specifically designed to select features that can significantly enhance the accuracy of classification. The proposed algorithm and competing ones, namely, Al-Biruni Earth Radius (BER), Particle Swarm Optimization (PSO), Whale Optimization Algorithm (WAO), Grey Wolf Optimizer (GWO) and Genetic Algorithm (GA) were evaluated using predefined sets of assessment criteria. The statistical analysis employed the ANOVA and Wilcoxon signed-rank tests and assessed the effectiveness and significance of the proposed algorithm compared to the other five algorithms. Furthermore, A series of visual depictions were presented to validate the effectiveness and robustness of the proposed algorithm. Thus, the MBER algorithm outperformed the other optimizers on the majority of the unimodal benchmark functions due to these considerations. Different ML models were used for classification, e.g., DT, RF, KNN, SGD, GNB, SVC, and LR. The KNN model achieved the highest values of Precision (PPV) (0.959425), Negative Predictive Value (NPV) (0.964969), FScore (0.963431), accuracy (0.9612), Sensitivity (0.970578) and Specificity (0.949711). Thus, KNN serves as a fitness function and is optimized by the utilization of Modified Al-Biruni earth radius (MBER). Finally, the accuracy of eye state classification achieved 96.12% using the proposed algorithm.

Determination of cyclic behavior of various cross-section composite column-beam connections reinforced with FRP composite using FEM analysis and ML models

ABSTRACT This study examined the cyclic behavior of glass-fiber-reinforced polymer composite column-beam connectors in various diameters. Mean learning (XGBoost and the random forest algorithm) and numerical methods were used to identify the data collected experimentally due to the rotation behavior. The initial sample had a column length of 1200 mm and a cross-section of 80 × 80 mm, with a beam section measuring 80 × 160 mm. The second sample had a 100 × 100 mm column and a beam section measuring 100 × 200 and 1200 mm, respectively. For the third sample, the column’s dimensions were 130 × 130 mm in cross-section and 1200 mm in length, and the beam was 130 × 260 mm. Spruce wood is used to make glulam components for columns and beams. Following the creation of the column-beam connections, glass-based fiber-reinforces polymer (FRP) fabric was used to wrap the column-beam connections that needed strengthening. The cyclic behavior of the reinforced reference samples and the samples with reinforced column-beam joint areas was examined. The column-beam connection with code C13B13-R has the most load-carrying capacity (13.9 kN), while the beam with code C08B08-UR has the lowest load-carrying capacity (03.4 kN). Examining the table reveals that the findings of the numerical analysis and the experiment provide comparable conclusions. When comparing experimental and numerical results, similar values are found for stiffness values (R 2 of 0.99), load-bearing capacity (R 2 of 0.99), and energy consumption capacity (R 2 of 0.99). With R 2 of 0.9978, RMSE of 0.01017, and MAE of 0.01043, the random forest approach identified the stiffness values in the model with the best accuracy. It has been shown that the seismic behavior of column-beam couplings built with these materials may be studied with mean learning models and numerical analysis, instead of demanding and time-consuming experimental studies.

Revealing Fibrosis Genes as Biomarkers of Ulcerative Colitis: A Bioinformatics Study Based on ScRNA and Bulk RNA Datasets.

This study aimed to uncover biomarkers associated with fibroblasts to diagnose ulcerative colitis (UC) and predict sensitivity to TNFα inhibitors. We identified fibrosis-related genes by analyzing eight bulk RNA and one single-cell RNA sequencing dataset from UC patients. Three machine learning algorithms were employed to identify common significant genes. We utilized five machine learning models, namely Random Forest (RF), Support Vector Machine (SVM), Xgboost, Multilayer Perceptron (MLP), and Logistic Regression, to develop diagnostic models for UC. Following hyperparameter tweaking using grid search, we evaluated Matthew's Correlation Coefficient (MCC) of each model on the validation set. Finally, we identified five hub genes in UC patients and evaluated their response to infliximab or golimumab. We identified 23 genes associated with fibroblasts. Further analysis using three ML models revealed BIRC3, IFITM2, ANXA1, ISG20, and MSN as critical fibroblast genes. Following hyperparameter adjustment, the SVM model exhibited the most favorable characteristics in the validation set, achieving an MCC of 0.7. ANXA1 contributed the most to the model that predicts UC. The optimal model was implemented on the website. Among UC patients receiving TNFα inhibitor treatment, the ineffective group showed considerably increased expression of the five critical genes than the responsive group. BIRC3, IFITM2, ANXA1, ISG20, and MSN may serve as potential diagnostic biomarkers in UC. Through the interaction between characteristic biomarkers and immune infiltrating cells, the immune response mediated by these characteristic biomarkers plays a crucial role in the occurrence and development of UC.

Data‐efficient graph learning: Problems, progress, and prospects

AbstractGraph‐structured data, ranging from social networks to financial transaction networks, from citation networks to gene regulatory networks, have been widely used for modeling a myriad of real‐world systems. As a prevailing model architecture to model graph‐structured data, graph neural networks (GNNs) have drawn much attention in both academic and industrial communities in the past decades. Despite their success in different graph learning tasks, existing methods usually rely on learning from “big” data, requiring a large amount of labeled data for model training. However, it is common that real‐world graphs are associated with “small” labeled data as data annotation and labeling on graphs is always time and resource‐consuming. Therefore, it is imperative to investigate graph machine learning (graph ML) with low‐cost human supervision for low‐resource settings where limited or even no labeled data is available. This paper investigates a new research field—data‐efficient graph learning, which aims to push forward the performance boundary of graph ML models with different kinds of low‐cost supervision signals. Specifically, we outline the fundamental research problems, review the current progress, and discuss the future prospects of data‐efficient graph learning, aiming to illuminate the path for subsequent research in this field.

A hybrid self-supervised model predicting life satisfaction in South Korea.

Life satisfaction pertains to an individual's subjective evaluation of their life quality, grounded in their personal criteria. It stands as a crucial cognitive aspect of subjective wellbeing, offering a reliable gauge of a person's comprehensive wellbeing status. In this research, our objective is to develop a hybrid self-supervised model tailored for predicting individuals' life satisfaction in South Korea. We employed the Busan Metropolitan City Social Survey Data in 2021, a comprehensive dataset compiled by the Big Data Statistics Division of Busan Metropolitan City. After preprocessing, our analysis focused on a total of 32,390 individuals with 51 variables. We developed the self-supervised pre-training TabNet model as a key component of this study. In addition, we integrated the proposed model with the Local Interpretable Model-agnostic Explanation (LIME) technique to enhance the ease and intuitiveness of interpreting local model behavior. The performance of our advanced model surpassed conventional tree-based ML models, registering an AUC of 0.7778 for the training set and 0.7757 for the test set. Furthermore, our integrated model simplifies and clarifies the interpretation of local model actions, effectively navigating past the intricate nuances of TabNet's standard explanatory mechanisms. Our proposed model offers a transparent understanding of AI decisions, making it a valuable tool for professionals in the social sciences and psychology, even if they lack expertise in data analytics.

Multi-method machine learning techniques in gold pathfinder elements prediction in central parts of Tanzania using stream sediment geochemical data

Prediction models using machine learning techniques have proven to be a reliable technique in mineral exploration. A combination of these techniques is very robust and reliable in exploration targeting and much dependable as an approach in greenfield. In this study, multi-machine learning methods: random forest (RF), support vector machine (SVM), and artificial neural network (ANN) were employed to conduct a data-driven gold (Au) prospectivity modelling in the Central parts of the Tanzania Craton (TC). A total of 166 samples with Au concentrations from stream sediment samples were considered. Based on the modeling results, the RF model demonstrates superior prediction accuracy compared to the SVM (MSE of 0.89) and ANN models (MSE of 1.21), achieving an MSE of less than 0.82. In terms of overall predictive performance and efficiency, the RF model outperforms other ML models deployed in this research. Therefore, it is deemed the suitable model for gold (Au) prediction in the TC catchments. According to the geological interpretation derived from the model, anomalies in arsenic (As), nickel (Ni), and tungsten (W) now emerge as significant predictors in the quest for gold. This implies that the association of As–Ni–W are potential pathfinder elements in the exploration of gold in the central part of the TC.

SAR-driven flood inventory and multi-factor ensemble susceptibility modelling using machine learning frameworks

Climate change has substantially increased both the occurrence and intensity of flood events, particularly in the Indian subcontinent, exacerbating threats to human populations and economic infrastructure. The present research employed novel ML models—LR, SVM, RF, XGBoost, DNN, and Stacking Ensemble—developed in the Python environment and leveraged 18 flood-influencing factors to delineate flood-prone areas with precision. A comprehensive flood inventory, obtained from Sentinel-1 Synthetic Aperture Radar (SAR) data using the Google Earth Engine (GEE) platform, provided empirical data for entire model training and validation. Model performance was assessed using precision, recall, F1-score, accuracy, and ROC-AUC metrics. The results highlighted Stacking Ensemble’s superior predictive ability (0.965), followed closely by, XGBoost (0.934), DNN (0.929), RF (0.925), LR (0.921), and SVM (0.920) respectively, establishing the feasibility of ML applications in disaster management. The maps depicting susceptibility to flooding generated by the current research provide actionable insights for decision-makers, city planners, and authorities responsible for disaster management, guiding infrastructural and community resilience enhancements against flood risks.

From online hate speech to offline hate crime: the role of inflammatory language in forecasting violence against migrant and LGBT communities

Social media messages often provide insights into offline behaviors. Although hate speech proliferates rapidly across social media platforms, it is rarely recognized as a cybercrime, even when it may be linked to offline hate crimes that typically involve physical violence. This paper aims to anticipate violent acts by analyzing online hate speech (hatred, toxicity, and sentiment) and comparing it to offline hate crime. The dataset for this preregistered study included social media posts from X (previously called Twitter) and Facebook and internal police records of hate crimes reported in Spain between 2016 and 2018. After conducting preliminary data analysis to check the moderate temporal correlation, we used time series analysis to develop computational models (VAR, GLMNet, and XGBTree) to predict four time periods of these rare events on a daily and weekly basis. Forty-eight models were run to forecast two types of offline hate crimes, those against migrants and those against the LGBT community. The best model for migrant crime achieved an R2 of 64%, while that for LGBT crime reached 53%. According to the best ML models, the weekly aggregations outperformed the daily aggregations, the national models outperformed those geolocated in Madrid, and those about migration were more effective than those about LGBT people. Moreover, toxic language outperformed hatred and sentiment analysis, Facebook posts were better predictors than tweets, and in most cases, speech temporally preceded crime. Although we do not make any claims about causation, we conclude that online inflammatory language could be a leading indicator for detecting potential hate crimes acts and that these models can have practical applications for preventing these crimes.

Prediction of urban surface water quality scenarios using hybrid stacking ensembles machine learning model in Howrah Municipal Corporation, West Bengal

In the pursuit of understanding surface water quality for sustainable urban management, we created a machine learning modeling framework that utilized Random Forest (RF), Cubist, Extreme Gradient Boosting (XGB), Multivariate Adaptive Regression Splines (MARS), Gradient Boosting Machine (GBM), Support Vector Machine (SVM), and their hybrid stacking ensemble RF (SE-RF), as well as stacking Cubist (SE-Cubist), to predict the distribution of water quality in the Howrah Municipal Corporation (HMC) area in West Bengal, India. Additionally, we employed the ReliefF and Shapley Additive exPlanations (SHAP) methods to elucidate the underlying factors driving water quality. We first estimated the water quality index (WQI) to model seven water quality parameters: total hardness (TH), pH, total dissolved solids (TDS), dissolved oxygen (DO), biochemical oxygen demand (BOD), calcium (Ca), magnesium (Mg). Then six independent factors were utilized (i.e. Precipitation (Pr), Maximum Temperature (Tmax), Minimum Temperature (Tmin), Normalized Difference Turbidity Index (NDTI), Normalized Difference Chlorophyll Index (NDCI), and Total Dissolved Solids (TDS)) for predicting the WQI mapping through the different ML models. This study demonstrated that the SE-Cubist model outperforms other ML models. During the testing phase, it achieved the best modeling results with an R2 = 0.975, RMSE = 0.351, and MAE = 0.197. The ReliefF and SHAP analyses identified Pr and Tmax as the most significant factors influencing WQI within the study area.

Global optimization: a machine learning approach

AbstractMany approaches for addressing global optimization problems typically rely on relaxations of nonlinear constraints over specific mathematical primitives. This is restricting in applications with constraints that are implicit or consist of more general primitives. Trying to address such limitations, Bertsimas and Ozturk (2023) proposed OCTHaGOn as a way of solving very general global optimization problems by approximating the nonlinear constraints using hyperplane-based decision-trees and then using those trees to construct a unified MIO approximation of the original problem. We provide extensions to this approach, by (i) approximating the original problem using other MIO-representable ML models besides decision trees, such as gradient boosted trees, multi layer perceptrons and suport vector machines (ii) proposing adaptive sampling procedures for more accurate ML-based constraint approximations, (iii) utilizing robust optimization to account for the uncertainty of the sample-dependent training of the ML models, (iv) leveraging a family of relaxations to address the infeasibilities of the final MIO approximation. We then test the enhanced framework in 81 global optimization instances. We show improvements in solution feasibility and optimality in the majority of instances. We also compare against BARON, showing improved optimality gaps and solution times in more than 9 instances.

A Hybrid Machine learning Techniques for Detection of Chronic Kidney Disease

Chronic kidney disease (CKD) is a prevalent medical condition that is distinguished by the progressive deterioration of renal function over time. The diagnosis, management, and treatment of this condition are exceedingly difficult due to its multifactorial etiology and extensive array of manifestations. The objective of this review paper is to provide a comprehensive examination of chronic kidney disease, including epidemiology, pathophysiology, clinical presentation, diagnostic criteria, treatment modalities, and preventive strategies. Health professionals must address the intricacies of the condition in order to manage or reduce the complications associated with CKD. Diabetes, hypertension, advanced age, and gender are among the numerous risk factors that increase the likelihood of developing CKD. The predictors of CKD were described in this investigation using machine learning algorithms. Our results indicate that ML can assist in the early identification and management of individuals who are at risk of developing CKD. The application of ML models to electronic health record data can reduce the prevalence of undiagnosed CKD, which is particularly beneficial in progressive diseases such as CKD, where the treatment burden and cost increase as the disease progresses.

Analyzing Machine Learning Models for Activity Recognition Using Homomorphically Encrypted Real-World Smart Home Datasets: A Case Study

The era of digitization and IoT devices is marked by the constant storage of massive amounts of data. The growing adoption of smart home environments, which use sensors and devices to monitor and control various aspects of daily life, underscores the need for effective privacy and security measures. HE is a technology that enables computations on encrypted data, preserving confidentiality. As a result, researchers have developed methodologies to protect user information, and HE is one of the technologies that make it possible to perform computations directly on encrypted data and produce results using this encrypted information. Thus, this research study compares the performance of three ML models, XGBoost, Random Forest, and Decision Classifier, on a real-world smart home dataset using both with and without FHE. Practical results demonstrate that the Decision Classifier showed remarkable results, maintaining high accuracy with FHE and even surpassing its plaintext performance, suggesting that encryption can enhance model accuracy under certain conditions. Additionally, Random Forest showed efficiency in terms of execution time and low prediction errors with FHE, making it a strong candidate for encrypted data processing in smart homes. These findings highlight the potential of FHE to set new privacy standards, advancing secure and privacy-preserving technologies in smart environments.

From Labquakes to Megathrusts: Scaling Deep Learning Based Pickers Over 15 Orders of Magnitude

AbstractThe application of machine learning techniques in seismology has greatly advanced seismological analysis, especially for earthquake detection and seismic phase picking. However, machine learning approaches still face challenges in generalizing to data sets that differ from their original training setting. Previous studies focused on retraining or transfer‐learning models for these scenarios, but require high‐quality labeled data sets. This paper demonstrates a new approach for augmenting already trained models without the need for additional training data. We propose four strategies—rescaling, model aggregation, shifting, and filtering—to enhance the performance of pre‐trained models on out‐of‐distribution data sets. We further devise various methodologies to ensemble the individual predictions from these strategies to obtain a final unified prediction result featuring prediction robustness and detection sensitivity. We develop an open‐source Python module quakephase that implements these methods and can flexibly process input continuous seismic data of any sampling rate. With quakephase and pre‐trained ML models from SeisBench, we perform systematic benchmark tests on data recorded by different types of instruments, ranging from acoustic emission sensors to distributed acoustic sensing, and collected at different scales, spanning from laboratory acoustic emission events to major tectonic earthquakes. Our tests highlight that rescaling is essential for dealing with small‐magnitude seismic events recorded at high sampling rates as well as larger magnitude events having long coda and remote events with long wave trains. Our results demonstrate that the proposed methods are effective in augmenting pre‐trained models for out‐of‐distribution data sets, especially in scenarios with limited labeled data for transfer learning.

Fractal dimension and lacunarity measures of glioma subcomponents are discriminative of the grade of gliomas and IDH status.

Since the overall glioma mass and its subcomponents-enhancing region (malignant part of the tumor), non-enhancing (less aggressive tumor cells), necrotic core (dead cells), and edema (water deposition)-are complex and irregular structures, non-Euclidean geometric measures such as fractal dimension (FD or "fractality") and lacunarity are needed to quantify their structural complexity. Fractality measures the extent of structural irregularity, while lacunarity measures the spatial distribution or gaps. The complex geometric patterns of the glioma subcomponents may be closely associated with the grade and molecular landscape. Therefore, we measured FD and lacunarity in the glioma subcomponents and developed machine learning models to discriminate between tumor grades and isocitrate dehydrogenase (IDH) gene status. 3D fractal dimension (FD3D) and lacunarity (Lac3D) were measured for the enhancing, non-enhancing plus necrotic core, and edema-subcomponents using preoperative structural-MRI obtained from the The Cancer Genome Atlas (TCGA) and University of California San Francisco Preoperative Diffuse Glioma MRI (UCSF-PDGM) glioma cohorts. The FD3D and Lac3D measures of the tumor-subcomponents were then compared across glioma grades (HGGs: high-grade gliomas vs. LGGs: low-grade gliomas) and IDH status (mutant vs. wild type). Using these measures, machine learning platforms discriminative of glioma grade and IDH status were developed. Kaplan-Meier survival analysis was used to assess the prognostic significance of FD3D and Lac3D measurements. HGG exhibited significantly higher fractality and lower lacunarity in the enhancing subcomponent, along with lower fractality in the non-enhancing subcomponent compared to LGG. This suggests that a highly irregular and complex geometry in the enhancing-subcomponent is a characteristic feature of HGGs. A comparison of FD3D and Lac3D between IDH-wild type and IDH-mutant gliomas revealed that mutant gliomas had ~2.5-fold lower FD3D in the enhancing-subcomponent and higher FD3D with lower Lac3D in the non-enhancing subcomponent, indicating a less complex and smooth enhancing subcomponent, and a more continuous non-enhancing subcomponent as features of IDH-mutant gliomas. Supervised ML models using FD3D from both the enhancing and non-enhancing subcomponents together demonstrated high-sensitivity in discriminating glioma grades (~97.9%) and IDH status (~94.4%). A combined fractal estimation of the enhancing and non-enhancing subcomponents using MR images could serve as a non-invasive, precise, and quantitative measure for discriminating glioma grade and IDH status. The combination of 2-hydroxyglutarate-magnetic resonance spectroscopy (2HG-MRS) with FD3D and Lac3D quantification may be established as a robust imaging signature for glioma subtyping.

ECBoA-OFS: An Ensemble Classification Model for Botnet Attacks based on Optimal Feature Selection using CPR in IoT

The rapid growth of the Internet of Things (IoT) has indeed introduced new security challenges, and the proliferation of compromised IoT devices has become a significant concern. Botnet attacks, where multiple corrupted devices are managed by a particular object, have become a widespread threat in IoT environments. These are used for a variety of malicious activities, including distributed DDoS attacks, data breaches, and malware distribution. However, detecting IoT botnets poses several challenges due to the resource constraints inherent in many IoT devices. The limitations in computation, storage, and communication capabilities make it challenging to deploy complex ML and deep learning models directly on these devices. This paper proposes an ensemble classification model ECBoA-OFS (Ensemble Classification for Botnet Attack Prediction using Optimal Feature Selection). It focuses on enhancing the accuracy of botnet attack prediction through the integration of ensemble methods and optimal feature selection. It describes a method for optimal feature selection in the context of analyzing the behavior of BoA and malicious traffic flow features in a network using Central Pivot Ranges (CPR). Feature selection is an important step in machine learning and data analysis because it supports to identification of the most important features for a given problem, thereby improving model performance and interpretation. The extracted features are used for model training and ensemble classification for prediction. To evaluate ECBoA-OFS, the N-BaIoT-2021 dataset consisting of regular IoT network traffic and BoA traffic records of corrupted IoT devices is utilized, considering detection precision, sensitivity, specificity, accuracy, and F1-score. Although all ensemble classifier models achieved better detection accuracy through optimal feature selection, the proposed ECBA-OFS shows better results compared to other ensemble classifier results.

Bias in Machine Learning: A Literature Review

Bias could be defined as the tendency to be in favor or against a person or a group, thus promoting unfairness. In computer science, bias is called algorithmic or artificial intelligence (i.e., AI) and can be described as the tendency to showcase recurrent errors in a computer system, which result in “unfair” outcomes. Bias in the “outside world” and algorithmic bias are interconnected since many types of algorithmic bias originate from external factors. The enormous variety of different types of AI biases that have been identified in diverse domains highlights the need for classifying the said types of AI bias and providing a detailed overview of ways to identify and mitigate them. The different types of algorithmic bias that exist could be divided into categories based on the origin of the bias, since bias can occur during the different stages of the Machine Learning (i.e., ML) lifecycle. This manuscript is a literature study that provides a detailed survey regarding the different categories of bias and the corresponding approaches that have been proposed to identify and mitigate them. This study not only provides ready-to-use algorithms for identifying and mitigating bias, but also enhances the empirical knowledge of ML engineers to identify bias based on the similarity that their use cases have to other approaches that are presented in this manuscript. Based on the findings of this study, it is observed that some types of AI bias are better covered in the literature, both in terms of identification and mitigation, whilst others need to be studied more. The overall contribution of this research work is to provide a useful guideline for the identification and mitigation of bias that can be utilized by ML engineers and everyone who is interested in developing, evaluating and/or utilizing ML models.