Diffusion-based generative models have demonstrated a capacity for perceptually
impressive synthesis, but can they also be great likelihood-based models? We answer this …

Score-based diffusion models synthesize samples by reversing a stochastic process that
diffuses data to noise, and are trained by minimizing a weighted combination of score …

It has long been established that predictive models can be transformed into lossless
compressors and vice versa. Incidentally, in recent years, the machine learning community …

Neural compression is the application of neural networks and other machine learning
methods to data compression. Recent advances in statistical machine learning have opened …

We review a class of methods that can be collected under the name nonlinear transform
coding (NTC), which over the past few years have become competitive with the best linear …

We introduce Autoregressive Diffusion Models (ARDMs), a model class encompassing and
generalizing order-agnostic autoregressive models (Uria et al., 2014) and absorbing …

Deep neural networks (DNNs) are often used for text classification due to their high
accuracy. However, DNNs can be computationally intensive, requiring millions of …

Implicit neural representations (INRs) recently achieved great success in image
representation and compression, offering high visual quality and fast rendering speeds with …

Lossless compression methods shorten the expected representation size of data without
loss of information, using a statistical model. Flow-based models are attractive in this setting …

This paper introduces Bayesian Flow Networks (BFNs), a new class of generative model in
which the parameters of a set of independent distributions are modified with Bayesian …