Artificial Intelligence Evolution

Comparison of the Artificial Neural Network's Approximations for the Levenberg-Marquardt Algorithm and the Gradient Descent Optimization on Datasets

Michael S. Osigbemeh, Chimereze Osuji, Moses O. Onyesolu, Uche P. Onochie — Mon, 11 Mar 2024 00:00:00 +0800

The approximations obtained by gradient descent optimization on a set of datasets were compared with the results obtained with the Levenberg-Marquardt Optimization Method (LMOM) on the same datasets. The datasets, which comprised three orthogonal databases obtained from MATLAB's Neural Network toolbox accompanying databases, were normalized and serially loaded to the artificial neural network Graphical User Interface (GUI) designed by the researchers. The GUI built with Visual Studio Programming Language (VSPL) implements a gradient descent optimization scheme of the back-propagation algorithm. The characteristics of each database for determination of the termination criteria were approximated from the developed feature extractive iteration algorithm. Revalidation sessions of the LMOM on the sampled datasets showed significant spuriousness in outputted results when compared with the gradient descent optimization results which although slow in attaining convergence produced results that can be closely replicated. Analysis of the F-statistics and the Receiver Operating Characteristics (ROC) for the sampled datasets results of both methods also showed that the gradient descent method demonstrated significant accuracy and parsimony in approximating the nonlinear solutions in the datasets when compared with the results from LMOM processing. Additionally, in this research, an algorithm for deducing and producing the ROC for analyzed Artificial Neural Network (ANN) sessions was also developed and implemented using VSPL.

MLMI: A Machine Learning Model for Estimating Risk of Myocardial Infarction

Subhagata Chattopadhyay — Thu, 25 Jan 2024 00:00:00 +0800

Cardiovascular diseases (CVD) are a global threat of high morbidity and mortality. Myocardial infarction (MI) due to coronary vessel malfunctions is one of the leading causes of mortality due to CVD. Interestingly, all CVD patients do not develop MI, and vice versa. Clinically, thus, it is a gray area. Therefore, an appropriate MI risk scoring (MIRS) tool could be useful to identify the high-risk (HR) population suffering from CVD. This research paper presents a hybrid machine learning (ML) model (MLMI) to identify MI risk where a) clustering of the CVD population with the help of the Gaussian mixture model (GMM) is used to identify the HR and not high-risk (NHR) groups, b) feature engineering of the members in both the HR and NHR populations using regression method that estimates the coefficient of determination (R²) to explore significant features to create the model by c) leveraging the R² values > 0.7 as the key features of the input dataset to a d) Feed-forward neural network (FFNN) for scoring the risk on a set of synthetic patient data, created by three experienced medical doctors. The myocardial infarction risk scores (MIRS) would assist users in prioritizing the patients needing monitoring and treatment. Finally, the MIRS values are validated by another group of three medical doctors to curb the research bias. The sensitivity, specificity, precision, F₁ scores, and accuracy of the MLMI model are computed to measure its efficiency. With limited input data, the proposed model shows an average accuracy, and precision of 77.33% each, while sensitivity and F₁ score are 100% and 88%, respectively.

Software Test Case Generation Using Natural Language Processing (NLP): A Systematic Literature Review

Halima Ayenew, Mekonnen Wagaw — Thu, 25 Jan 2024 00:00:00 +0800

Technologies for natural language processing (NLP) are employed to assist in the analysis and comprehension of human language. Researchers are increasingly focusing on NLP techniques to automate various software development tasks, such as software testing (test case generation). However, choosing the best NLP methods to create automated test cases is never simple. As a result, we look into using NLP techniques to create test cases. We identified 13 research articles published between 2015 and 2023 for this study. As a result, to generate automated test cases, 7 NLP techniques, 2 tools, and 1 framework have been suggested. In addition, 7 NLP algorithms have been discovered in the context of test case generation. Our evaluations indicate that the identified NLP techniques are very useful for automating the generation of test cases. The successful completion of software testing processes (test case generation) therefore requires the use of this approach/technique by software developers, testers, and software engineering teams in general. This paper will be beneficial for researchers engaged in the automation of software testing. Furthermore, it will also be helpful for academic researchers and software engineers (testers) seeking insights into the state of the art in test case generation automation. The paper discusses various tools and methods proposed for test case generation automation, aiding readers in evaluating and selecting the most suitable method for automated test case generation.