In contrast to V AE-based neural compression, where the (mean) decoder is a deterministic neural network, our decoder is a conditional diffusion model. Our approach thus introduces an additional "content" latent variable on which the reverse diffusion process

Many practical settings call for the reconstruction of temporal signals from corrupted or missing data. Classic examples include decoding, tracking, signal enhancement and denoising.

Our algorithm is few-shot: Given about a dozen images restored by the model, it can significantly improve the perceptual quality and/or the MSE of the model for newly restored images without further training.

Image restoration tasks are ill-posed problems whose solutions are not unique.

Neural Information Processing Systems http://nips.cc/

In the Gaussian setting, we further provide a closed form expression for such estimators.

The pursuit of high perceptual quality in image restoration has driven the development of revolutionary generative models, capable of producing results often visually indistinguishable from real data. However, as their perceptual quality continues toimprove, these models also exhibit agrowing tendencytogenerate hallucinations -realistic-looking details that do not exist in the ground truth images. Hallucinations in these models create uncertainty about their reliability, raising major concerns about their practical application.

We prove that the corresponding information-theoretic universal rate-distortionperception function is operationally achievable in an approximate sense.

The internet age relies on lossy compression algorithms that transmit information at low bitrates. These algorithms are typically analysed through the rate-distortion trade-off, originally posited by Shannon[33].