Abstract
Autoencoders have become a fundamental technique in deep learning (DL), significantly enhancing representation learning across various domains, including image processing, anomaly detection, and generative modelling. This paper provides a comprehensive review of autoencoder architectures, from their inception and fundamental concepts to advanced implementations such as adversarial autoencoders, convolutional autoencoders, and variational autoencoders, examining their operational mechanisms, mathematical foundations, typical applications, and their role in generative modelling. The study contributes to the field by synthesizing existing knowledge, discussing recent advancements, new perspectives, and the practical implications of autoencoders in tackling modern machine learning (ML) challenges.