Understanding Taxonomy of Generative Models

Abstract

The study of generative models has gained popularity in the field of machine learning. These models are valuable for a range of applications, including picture and speech synthesis, text generation, and anomaly detection, since they can produce new data that is similar to the training data. We give a thorough overview of the many generative model types in this review paper, covering pixelCNN/RNN, variational autoencoders and generative adversarial networks but the main focus is on GAN nets. We compare each model's performance on a variety of tasks and talk about its advantages, disadvantages and applications. In conclusion, this review paper offers a thorough and current summary of the state-of-the-art in generative modeling and should be helpful for scholars and practitioners interested in this fascinating and quickly developing topic.