Deep Learning Architectures for Accurate Brain Tumour Analysis

Abstract

Machine learning is concerned with using computers to construct and develop learning algorithms. Handwriting recognition, biometric recognition, stock market analysis, and medical diagnosis are some of the uses. The algorithms are divided into supervised (learning from training examples) and unsupervised (learning from random samples) (a model is fit to observations). Deep learning involves the use of intricate models in order to acquire knowledge from a substantial amount of data, such as observations or images. It might be supervised (image classification) or unsupervised (non-supervised) (image compression). A variety of deep learning-based research initiatives have been established to increase the diagnostic accuracy of brain cancer classification. This paper covers the brain imaging datasets utilized in several brain image classification systems and describes the generally used metrics for assessing the systems' performance. There is also information on various deep-learning algorithms, such as the number of tumor classes and samples.