This chapter provides a detailed introduction to unsupervised learning, supervised learning, and semi-supervised learning, including the definitions, essences, common scenarios, and frequently used models of different learning methods. Subsequently, generative models and discriminative models are respectively introduced respectively within the scope of supervised learning, covering their definitions, differences, common models, etc. Then, the concept and learning approach of unsupervised generative models are presented. In the second part of this chapter, we classify generative models into two types, explicit generative models and implicit generative models, according to the way generative models handle the probability density function. For explicit generative models, the principle of the maximum likelihood method is described in detail and is divided into two categories: tractable probability density functions and approximate methods. In the first category, FVBN series models are listed, including PixelRNN, PixelCNN, NADE, and flow models. In the second category, variational autoencoders and restricted Boltzmann machines are introduced. In the third part, the implicit generative model is introduced taking GAN as an example, and GAN is compared with other generative models.