Learned transform compression with optimized entropy encoding
This work addresses compression efficiency for data storage and transmission, presenting a novel approach that could improve performance in this domain.
The paper tackles the problem of learned transform compression by jointly learning the transform and probability distribution of discrete codes, using soft relaxations for quantization and entropy optimization, and demonstrates its effectiveness through proof-of-concept experiments.
We consider the problem of learned transform compression where we learn both, the transform as well as the probability distribution over the discrete codes. We utilize a soft relaxation of the quantization operation to allow for back-propagation of gradients and employ vector (rather than scalar) quantization of the latent codes. Furthermore, we apply similar relaxation in the code probability assignments enabling direct optimization of the code entropy. To the best of our knowledge, this approach is completely novel. We conduct a set of proof-of concept experiments confirming the potency of our approaches.