Evaluation and Comparison of Machine Learning Algorithms for Effective Image Classification with Fault-Tolerance

Machine learning today becomes more and more effective instrument to solve many particular problems, where there are difficulties to apply well known and described math model. In other words - it is a great tool to describe nonlinear phenomena. We tried to use this technique to improve existing process of stratigraphy and lithology interpretation and reduce costs on site by applying computer leaded predictions on the basis of existing on-field collected data. Article describes usage of machine learning algorithms for several geology data stratigraphy and lithology boundaries classification based on geophysics logging data for deposits in Kazakhstan. Correct marking of stratigraphy and lithology from geophysics logging data is complex non-linear task. To solve this task we applied several algorithms of machine learning: random forest, logistic regression, gradient boosting (scikit-learn library), k - nearest neighbour (KNN) and XGBoost.

Remote cameras have become a widespread data‐collection tool for terrestrial mammals, but classifying images can be labor intensive and limit the usefulness of cameras for broad‐scale population monitoring. Machine learning algorithms for automated image classification can expedite data processing, but image misclassifications may influence inferences. Here, we used camera data for three sympatric species with disparate body sizes and life histories – black‐tailed jackrabbits (Lepus californicus), kit foxes (Vulpes macrotis), and pronghorns (Antilocapra americana) – as a model system to evaluate the influence of competing image classification approaches on estimates of occupancy and inferences about space use. We classified images with: (i) single review (manual), (ii) double review (manual by two observers), (iii) an automated‐manual review (machine learning to cull empty images and single review of remaining images), (iv) a pretrained machine‐learning algorithm that classifies images to species (base model), (v) the base model accepting only classifications with ≥95% confidence, (vi) the base model trained with regional images (trained model), and (vii) the trained model accepting only classifications with ≥95% confidence. We compared species‐specific results from alternative approaches to results from double review, which reduces the potential for misclassifications and was assumed to be the best approximation of truth. Despite high classification success, species‐level misclassification rates for the base and trained models were sufficiently high to produce erroneous occupancy estimates and inferences related to space use across species. Increasing the confidence thresholds for image classification to 95% did not consistently improve performance. Classifying images as empty (or not) offered a reasonable approach to reduce effort (by 97.7%) and facilitated a semi‐automated workflow that produced reliable estimates and inferences. Thus, camera‐based monitoring combined with machine learning algorithms for image classification could facilitate monitoring with limited manual image classification.

This paper presents a comparative study on machine learning algorithms for neutral section image classification. The classifiers are trained by employing the Histogram of Oriented Gradient features that are extracted from the neutral section dataset [1]. A neutral section is a phase break that is used on the Transnet freight rail system to separate the single-phase supply from the 25kV three-phase overhead traction supply. The 25kV is a stepped-down voltage from an 8SkV three-phase supply coming from the national grid. While the main purpose of the neutral section is to separate phase voltages, electric locomotives can traverse through these phases by switching On and Off. This auto-switching is possible through induction magnets installed in between the rails and with magnet detection sensors installed underneath the locomotives. However, a computer vision model has been developed, trained, and tested with a neutral section dataset containing images having open and close markers [1]. This paper, therefore, utilises this dataset to provide performance comparison on several machine learning classification algorithms viz. Decision Tree, Discriminant Analysis, Support Vector Machine, K-Nearest Neighbors, Ensemble, Naïve Bayes, and Convolutional Neural Network. A confusion matrix, F1-measure and computation time are employed to measure the performance of each classifier. The MATLAB Classification Learner application was used to obtain the results. The results show that the Linear Support Vector Machine performs best when considering performance and prediction speed. The Linear Support Vector Machine achieved a training accuracy of 93.40% with a test accuracy reaching 94% at a prediction speed of 75 objects per second (computation time).

Classification is a widely researched area in the machine learning and fuzzy communities with several approaches proposed by both communities. Some of the most relevant rule-based approaches from the machine learning community might include decision trees and rule inducers. The fuzzy community has also proposed many rule-based approaches, such as fuzzy decision trees and genetic fuzzy systems. This paper aims at comparing the models generated by rule-based methods for classification from both communities in terms of accuracy, and the induced rule set in terms of the syntactic complexity, taking into account the number of rules and average number of conjunctions in those rules. In general, models with lower syntactic complexity also show better interpretability, which is an important issue in knowledge acquisition. Results, using 10 datasets, a fuzzy C4.5 algorithm and two classic machine learning algorithms (C4.5 and PART), show that the fuzzy approach is able to produce lower error rates. Regarding the syntactic complexity of the models, PART produces in most cases the simplest models, although learning from different sets of features selected by filters. However, these simple models do not necessarily show a low error rate. Nevertheless, the induced fuzzy models inherit, from the fuzzy logic, the embedded ability of processing uncertainty and imprecision, avoiding the creation of rules using unnatural divisions of the attributes as the classic algorithms might do.

Advancements in the implementation of quantum hardware have enabled the acquisition of data that are intractable for emulation with classical computers. The integration of classical machine learning (ML) algorithms with these data holds potential for unveiling obscure patterns. Although this hybrid approach extends the class of efficiently solvable problems compared to using only classical computers, this approach has been only realized for solving restricted problems because of the prevalence of noise in current quantum computers. Here, we extend the applicability of the hybrid approach to problems of interest in many-body physics, such as predicting the properties of the ground state of a given Hamiltonian and classifying quantum phases. By performing experiments with various error-reducing procedures on superconducting quantum hardware with 127 qubits, we managed to acquire refined data from the quantum computer. This enabled us to demonstrate the successful implementation of theoretically suggested classical ML algorithms for systems with up to 44 qubits. Our results verify the scalability and effectiveness of the classical ML algorithms for processing quantum experimental data.

Thyroid disorders, affecting millions of individuals across the globe, require prompt and reliable diagnosis for optimal treatment and better patient results. On the other hand, conventional diagnostic tools are usually time-consuming and human-biased. This paper reviews an exploratory comparison of several machine learning (ML) algorithms for early diagnosis and classification of thyroid diseases based on their ability to automatize and hence the medical diagnosis. Through the comparison of the strengths and weaknesses of various ML methods, we assess them in terms of accuracy, precision, F1 score, and their applicability to clinical use. Our study utilizes datasets containing thyroid-related factors such as age, gender, TSH, T3 followed by feature selection and compares the performance of various ML techniques for thyroid disease. The purpose of this study is to contribute to the expanding literature on how machine learning can be effectively used for diagnosis enhancement of thyroid diseases and classify it into: hypothyroid, hyperthyroid, euthyroid.

Machine Learning Algorithms for Stratigraphy Classification on Uranium Deposits

In this work, we studied the use of Quantum Machine Learning (QML) algorithms for binary classification and compared their performance with classical Machine Learning (ML) methods. QML merges principles of Quantum Computing (QC) and ML, offering improved efficiency and potential quantum advantage in data-driven tasks and when solving complex problems. In binary classification, where the goal is to assign data to one of two categories, QML uses quantum algorithms to process large datasets efficiently. Quantum algorithms like Quantum Support Vector Machines (QSVM) and Quantum Neural Networks (QNN) exploit quantum parallelism and entanglement to enhance performance over classical methods. This study focuses on two common QML algorithms, Quantum Support Vector Classifier (QSVC) and QNN. We used the Qiskit software and conducted the experiments with three different datasets. Data preprocessing included dimensionality reduction using Principal Component Analysis (PCA) and standardization using scalers. The results showed that quantum algorithms demonstrated competitive performance against their classical counterparts in terms of accuracy, while QSVC performed better than QNN. These findings suggest that QML holds potential for improving computational efficiency in binary classification tasks. This opens the way for more efficient and scalable solutions in complex classification challenges and shows the complementary role of quantum computing.

Deep learning is a subfield of machine learning and deep neural architectures can extract high level features automatically without handcraft feature engineering unlike traditional machine learning algorithms. In this paper, we propose a method, which combines feature extraction layers of a convolutional neural network with traditional machine learning algorithms, such as, support vector machine, gradient boosting machines, and random forest. All of the proposed hybrid models and the above mentioned machine learning algorithms are trained on three different datasets: MNIST, Fashion-MNIST, and CIFAR-10. Results show that the proposed hybrid models are more successful than traditional models while they are being trained from raw pixel values. In this study, we empower traditional machine learning algorithms for classification using feature extraction ability of deep neural network architectures and we are inspired by transfer learning methodology to this.

In brain–computer interfaces (BCIs), it is crucial to process brain signals to improve the accuracy of the classification of motor movements. Machine learning (ML) algorithms such as artificial neural networks (ANNs), linear discriminant analysis (LDA), decision tree (D.T.), K-nearest neighbor (KNN), naive Bayes (N.B.), and support vector machine (SVM) have made significant progress in classification issues. This paper aims to present a signal processing analysis of electroencephalographic (EEG) signals among different feature extraction techniques to train selected classification algorithms to classify signals related to motor movements. The motor movements considered are related to the left hand, right hand, both fists, feet, and relaxation, making this a multiclass problem. In this study, nine ML algorithms were trained with a dataset created by the feature extraction of EEG signals.The EEG signals of 30 Physionet subjects were used to create a dataset related to movement. We used electrodes C3, C1, CZ, C2, and C4 according to the standard 10-10 placement. Then, we extracted the epochs of the EEG signals and applied tone, amplitude levels, and statistical techniques to obtain the set of features. LabVIEW™2015 version custom applications were used for reading the EEG signals; for channel selection, noise filtering, band selection, and feature extraction operations; and for creating the dataset. MATLAB 2021a was used for training, testing, and evaluating the performance metrics of the ML algorithms. In this study, the model of Medium-ANN achieved the best performance, with an AUC average of 0.9998, Cohen’s Kappa coefficient of 0.9552, a Matthews correlation coefficient of 0.9819, and a loss of 0.0147. These findings suggest the applicability of our approach to different scenarios, such as implementing robotic prostheses, where the use of superficial features is an acceptable option when resources are limited, as in embedded systems or edge computing devices.

Although the consortium blockchains commonly adopt consensus algorithms with Byzantine fault tolerance (such as practical Byzantine fault tolerance (PBFT)), its consensus efficiency will be degraded by the existence of malicious nodes or behaviors. However, the existing researches mainly focus on the detection of malicious behaviors for public blockchains, but are rare about consortium blockchains. In this paper, an anomaly detection model based on machine learning (ML) classification algorithm is proposed for consortium blockchains that adopt PBFT. Besides, a two-stage process is proposed to reduce the resource consumption for anomaly detection. The data needed for proposed model only has two dimensions and is convenient to obtain. The results of experiment show that ML is very effective in anomaly detection for consortium blockchains. Specifically, the algorithms with the highest accuracy are convolutional neural networks (CNN), k-nearest neighbor (KNN) and support vector machines (SVM) in turn. However, KNN and SVM are more suitable because resource consumption of both algorithms are one third of CNN, and the accuracy rates are above 0.9 which is 0.9% lower than CNN.

Despite the promise for performance and accuracy improvements of quantum inspired (QI) algorithms over classical machine learning (ML) algorithms, such gains have not been realized in practice. The quantum inspired algorithms can theoretically achieve significant speed up based on sampling assumptions and have thus far failed to outperform the existing classical ML models in practical applications. The speedup of quantum machine learning (QML) algorithms assume the access to data in quantum random access memory (QRAM) which is a strong assumption with current quantum architectures. QI algorithms assume sample and query (SQ) access to input vector and norms of matrices using a dynamic data structure. We explore the components of these models and the assumptions in this paper by surveying the recent works in QML and QI Machine learning (QIML) algorithms. We limit our study to QML and QIML models on achieving a speed up over classical ML techniques rather than individual proofs of these algorithms. This study highlights the assumptions being made that are currently not practical for QML and QIML algorithms in achieving performance advantage over classical ML algorithms.

Epilepsy is a serious brain illness characterized by abnormal brain activity. To build a better computer-aided diagnosis (CAD) solution, we provide a comprehensive comparison of four prominent Machine Learning (ML) algorithms for epilepsy classification: Naive Bayes (NB), K-Nearest Neighbor (K-NN), Decision Tree (DT), and Logistic Regression (LR). One of the goals of the proposed model was to address current challenges such as outliers, low performance, and to build a stable and accurate classification model. As a result, few notable sampling techniques are employed to address the problems associated with the dataset. Cohen's distribution class's Time-frequency domain (TFD) technique is used to extract significant properties from the EEG dataset. Following the retrieval of the sample data, different algorithms were used to retain the most suitable features from the derived data, such as Univariate Logistic Regression based feature selection (ULOGR), Gini-Score based significant feature selection (GSFR), and Information Gain based features (IGFR). Following the classification, the performance of classifiers was compared before and after the addition of feature selection methods to the two datasets using the different t-f methods utilized in this study. Our findings suggest that the performance of the classifier improves i) after dimensionality reduction ii) if the t-f technique employs frequency and time smoothing windows. For the UCI Dataset, the obtained findings reveal that the NB classifier excels the other classifiers with a 96% accuracy rate.

In the realm of Internet of Things (IoT) security, malware classification is crucial for identifying, isolating, and mitigating the impacts of malicious software that exploits system vulnerabilities. Effective IoT malware classification employs various machine learning algorithms to enhance detection accuracy and system resilience. The most commonly used Machine Learning algorithms for Malware Classification are "Support Vector Machine (SVM), Random Forest (RF), Naive Bayes (NB), and Logistic Regression (LR)". Each of these algorithms has distinct advantages in identifying malware, but this study explores the creation of an Ensemble / Hybrid model by integrating the two most effective machine learning algorithms, which can lead to superior performance. The proposed model outperforms the simple models mentioned above and other related works discussed in the literature. Key performance metrics evaluated for this work includes Accuracy, Precision, Recall, F-Score. This advancement is vital for improving the security posture of IOT systems by providing a more reliable mechanism for early detection and effective response to emerging malware threats

Active breeding programs specifically for root system architecture (RSA) phenotypes remain rare; however, breeding for branch and taproot types in the perennial crop alfalfa is ongoing. Phenotyping in this and other crops for active RSA breeding has mostly used visual scoring of specific traits or subjective classification into different root types. While image-based methods have been developed, translation to applied breeding is limited. This research is aimed at developing and comparing image-based RSA phenotyping methods using machine and deep learning algorithms for objective classification of 617 root images from mature alfalfa plants collected from the field to support the ongoing breeding efforts. Our results show that unsupervised machine learning tends to incorrectly classify roots into a normal distribution with most lines predicted as the intermediate root type. Encouragingly, random forest and TensorFlow-based neural networks can classify the root types into branch-type, taproot-type, and an intermediate taproot-branch type with 86% accuracy. With image augmentation, the prediction accuracy was improved to 97%. Coupling the predicted root type with its prediction probability will give breeders a confidence level for better decisions to advance the best and exclude the worst lines from their breeding program. This machine and deep learning approach enables accurate classification of the RSA phenotypes for genomic breeding of climate-resilient alfalfa.