Adaptive weighing of context models for lossless data compression

By Matthew V. Mahoney. 2005

169 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Information Theory

Citations in the corpus

Programmable Agents (2017, neutral): “Use of programs to specify hard deterministic gating is also reminiscent of PAQ8 [24, 18].”
SUBWORD LANGUAGE MODELING WITH NEURAL NETWORKS (2011, methodology): “Most statistical language models use words rather than char a ters as their atomic units due to their superior performance in term s of accuracy and the number of parameters to be estimated [1].”
A Machine Learning Perspective on Predictive Coding with PAQ (2011, methodology): “To the best of our knowledge, there exist only incomplete high-level descriptions of PAQ1 through 6 (Mahoney, 2005) and PAQ8 (Mahoney, accessed April 15, 2011).”
Online Learning with Gated Linear Networks (2017, background): “…the PAQ8 variant (Mahoney, 2013) from a machine learning perspective, providing a useful interpretation of the as a kind of mixture of experts (Jacobs et al., 1991) architecture, as well as providing an up-to-date summary of the algorithmic developments since the tech report of Mahoney (2005).”
A Machine Learning Perspective on Predictive Coding with PAQ8 (2012, methodology): “To the best of our knowledge, there exist only incomplete high-level descriptions of PAQ1 through 6 [3] and PAQ8 [6].”
Generating Text with Recurrent Neural Networks (2011, methodology): “While our method performs at the state of the art for pure character-level models, its compression performance falls short of the best models which have explicit knowledge of words, the most powerful of these being PAQ8hp12 (Mahoney , 2005).”