This paper provides a comprehensive survey of autoencoders, covering their principles, development, and applications in various fields. Autoencoders are unsupervised learning models that learn data features and reduce dimensionality. The authors present a taxonomy of autoencoders based on their structures and principles, analyzing and discussing related models. They review the applications of autoencoders in machine vision, natural language processing, complex networks, recommender systems, speech processing, anomaly detection, and other areas. The paper also discusses the limitations of current autoencoder algorithms and outlines future research directions. Key contributions include a new taxonomy of autoencoder methods, detailed explanations of various autoencoder architectures, and a review of their applications and software platforms. The authors aim to address three main research questions: the types of AE algorithms, methodological frameworks, and future directions in the field. The paper highlights the versatility and importance of autoencoders in machine learning and data analysis, emphasizing their role in dimensionality reduction, feature extraction, and various downstream tasks.This paper provides a comprehensive survey of autoencoders, covering their principles, development, and applications in various fields. Autoencoders are unsupervised learning models that learn data features and reduce dimensionality. The authors present a taxonomy of autoencoders based on their structures and principles, analyzing and discussing related models. They review the applications of autoencoders in machine vision, natural language processing, complex networks, recommender systems, speech processing, anomaly detection, and other areas. The paper also discusses the limitations of current autoencoder algorithms and outlines future research directions. Key contributions include a new taxonomy of autoencoder methods, detailed explanations of various autoencoder architectures, and a review of their applications and software platforms. The authors aim to address three main research questions: the types of AE algorithms, methodological frameworks, and future directions in the field. The paper highlights the versatility and importance of autoencoders in machine learning and data analysis, emphasizing their role in dimensionality reduction, feature extraction, and various downstream tasks.