I think that the title of the paper should state that it is for lossy image compression, which clearly states how it works and what task it performs. I would b

I think that the title of the paper should state that it is for lossy image compression, which clearly states how it works and what task it performs. I would be surprised if there wasn't a way to provide a learned, lossless method of compression, but that would be a very different paper and result.