Arbitrary source models and Bayesian codebooks in rate-distortion theory

The development of a universal lossy data compression model based on a lossy version of the Kraft inequality was presented. The performance of universal mixture codebooks was presented as an application of the results. The compression perforemance of codes and their distortion level were defined. The lossy Kraft inequality, non asymptotic bounds and pointwise redundancy and the theorems relating to the redundancy of codes generated by mixture codebooks or Bayesian codebooks were proved