In the new paper Neural Diffusion Processes, a research team from the University of Cambridge, Secondmind, and Google Research presents Neural Diffusion Processes (NDPs), a novel framework that learns to sample from rich distributions over functions at a lower computational cost than the true Bayesian posterior of a conventional Gaussian process.
In applying the adversarial training, this paper adopts distributed word representation, or word embedding, as the input, rather than the traditional one-hot representation. The reason lies in the fact that the higher dimensionality the input has, the more likely it is to be disturbed by noise.
Word2vec is an open source tool developed by a group of Google researchers led by Tomas Mikolov in 2013. It describes several efficient ways to represent words as M-dimensional real vectors, also known as word embedding, which is of great importance in many natural language processing applications