Machine Learning Approaches for Predicting AIDS Virus Infection: A Comparative Study using Random Forest and Neural Networks

Machine Learning Approaches for Predicting AIDS Virus Infection: A Comparative Study using Random Forest and Neural Networks

Amrutha, Ananya S Padasalgi, Maanasa M G, Smrithi R Holla, Dr. Shivandappa

Biotechnology
R V College of Engineering
Bangalore , India

Abstract—This review investigates the use of machine learning approaches, notably Random Forest and Neural Network classifiers, in the context of AIDS classification and digit identification using the MNIST dataset. The paper compares the performance of a Random Forest classifier and a Multi-Layer Perceptron (MLP) neural network on an AIDS classification dataset, emphasizing the significance of feature scaling and the impact of model design on classification accuracy. The Random Forest model was used to determine feature relevance, and the MLP classifier was trained and tested for accuracy in categorizing the binary outcome of HIV infection.

Furthermore, the study presents a proprietary neural network model built to identify handwritten digits from the MNIST dataset, emphasizing the model's architecture, which includes dense layers with ReLU activation and dropout regularization to reduce overfitting. The models were trained and tested, with performance indicators including accuracy and loss measured across epochs.

The results reveal that both traditional machine learning approaches and deep learning architectures can handle various datasets, highlighting their capabilities in binary classification and multiclass digit identification tasks. The results highlight the potential for integrating feature selection techniques with advanced neural network models to improve predicted performance in medical diagnostics and picture categorization.

Keywords—machine learning, formatting, google colab, neural network