Some results on adaptive statistics estimation for the reversible compression of sequences

Extended summary form only given. This paper is concerned with statistics estimation methods and how they influence model performance. For simplicity, a memoryless model is assumed with a single context and binary alphabet. Nonadaptive estimation uses the relative frequency counts of the letters (0 and 1) in the entire sequence. Depletive adaptive estimation uses the relative frequency counts of the letters in the part of the sequence yet to be scanned. Cumulative adaptive estimation uses the relative frequency counts in the part of the sequence already scanned