On the compression dimension of data strings and data sets

John C. Kieffer, G. Nelson · 2002

A hierarchical lossless source code compresses data by means of a graph used to represent the data. We show that the hierarchical codes which perform best as the number of data samples grows have a compression performance that can be characterized via a notion of the dimension of the data which we call the compression dimension.

Read the paper · More papers on PaperTik