Variational Mixture of HyperGenerators for Learning Distributions Over Functions
This addresses the need for efficient inference in generative models for continuous functions, applicable to domains like images and climate data, but it appears incremental as it builds on existing methods like INRs and VAEs.
The paper tackled the problem of computationally costly inference tasks in generative models over function spaces by proposing VAMoH, a model combining implicit neural representations and variational autoencoders, which achieved effective learning of distributions over continuous functions and performed inference tasks like conditional super-resolution and in-painting as well or better than previous approaches with less computational demand.
Recent approaches build on implicit neural representations (INRs) to propose generative models over function spaces. However, they are computationally costly when dealing with inference tasks, such as missing data imputation, or directly cannot tackle them. In this work, we propose a novel deep generative model, named VAMoH. VAMoH combines the capabilities of modeling continuous functions using INRs and the inference capabilities of Variational Autoencoders (VAEs). In addition, VAMoH relies on a normalizing flow to define the prior, and a mixture of hypernetworks to parametrize the data log-likelihood. This gives VAMoH a high expressive capability and interpretability. Through experiments on a diverse range of data types, such as images, voxels, and climate data, we show that VAMoH can effectively learn rich distributions over continuous functions. Furthermore, it can perform inference-related tasks, such as conditional super-resolution generation and in-painting, as well or better than previous approaches, while being less computationally demanding.