… rate-distortion autoencoders.Here we will describe the specific models we use for encoder,
code model, and decoder, as well as the data format, preprocessing, and loss functions. …

… ratedistortion theory, … rate depends not only on the distortion, but also on the perceptual
quality of the algorithm. A preliminary attempt to incorporate perceptual quality into rate-distortion …

… DILATE (DIstortion Loss … to models trained with the standard Mean Squared Error (MSE)
loss function, and also to DTW and variants. DILATE is also agnostic to the choice of the model, …

… The coding improvements provided by CC and LRP are most effective at low bit rates where
our model saves more than 16% compared to the context-adaptive baseline and as much …

… For learned image compression, the autoregressive context model is proved effective in
improving the ratedistortion (RD) performance. Because it helps remove spatial redundancies …

… However, we haven’t observed this effect for image compression models and practically
interesting rate–distortion trade-offs, suggesting that this is only the case for extremely high …

… As we can see from the results, great progress has been made to improve the ratedistortion
performance, where the decorrelation normalization and the hyperprior model bring …

… To estimate the entropy we will use the entropy model described in [8] to approximate Pq
by pz(z). Finally, we will use mean squared error (MSE) as a distortion metric. With these …

… ratedistortion … the ratedistortion performance regardless of the model capacity. Besides,
GMM works better on 192 filters than 128 filters, probably because 192 filters have large model …

… We observe that the model learns to compensate for complex motion in crowded scenes,
predicting flow-like displacement fields while purely being trained for the rate–distortion …