A novel approach of machine learning application in astrophysics: morphological feature wrapping based ensemble method for galaxy shape classification using GAMA dataset

The numerous strategies for the automated morphological categorization of galaxies, which uses a variety of supervised machine learning techniques, have not been well examined or compared. As the majority of star galaxy classifiers in use today use condensed summary data from catalogues, rigorous feature extraction and selection are required. With the aid of Deep Convolutional Neural Networks (CNN), a development in machine learning, it may automate the process of feature detection from data by a computer, therefore lowering the demand for qualified human input. Low-level artificial classification has made great progress. While this is the case, Deep Learning consistently outperforms traditional computers. analyzing large datasets while learning. We examine three machine learning techniques for categorizing morphological galaxies: Support Vector Machines (SVM), Random Forests (RF), and Naive Bayes (NB). We examine the efficacy of several machine learning algorithms on each feature representation of a galaxy using a collection of morphological features produced by image analysis as well as the raw image pixel data compressed using PCA (Principal Component Analysis) into PCA features. According to our experiments, RF outperformed SVM and NB. The remainder of the time, morphological features outperformed our PCA features in performance. Thus, the current mechanism is not extremely scalable. A probabilistic classifier that can scale, is based on source data, and requires the least amount of human interaction is essential to resolving these problems