Destruction is a General Strategy to Learn Generation; Diffusion's Strength is to Take it Seriously; Exploration is the Future
This paper provides a conceptual framework for understanding diffusion models and suggests future research directions for the machine learning community, particularly concerning exploration in diffusion-native ways.
This paper explores diffusion models as a method of learning generation by withholding information from a model's input and training it to guess the withheld information. It argues that diffusion's destructive approach is more flexible than hand-crafted techniques, especially in data-scarce environments.
I present diffusion models as part of a family of machine learning techniques that withhold information from a model's input and train it to guess the withheld information. I argue that diffusion's destroying approach to withholding is more flexible than typical hand-crafted information withholding techniques, providing a rich training playground that could be advantageous in some settings, notably data-scarce ones. I then address subtle issues that may arise when porting reinforcement learning techniques to the diffusion context, and wonder how such exploration problems could be addressed in more diffusion-native ways. I do not have definitive answers, but I do point my fingers in directions I deem interesting. A tutorial follows this thesis, expanding on the destroy-then-generate perspective. A novel kind of probabilistic graphical models is introduced to facilitate the tutorial's exposition.