Adaptive Context Modeling for Arithmetic Coding Using Perceptrons

Arithmetic coding is used in most media compression methods. Context modeling usually done through frequency counting and look-up tables (LUTs). For long-memory signals, probability with large context sizes often infeasible. Recently, neural networks have been to model probabilities of contexts order drive arithmetic coders. These trained offline. We introduce an online method for training a perceptron-based context-adaptive coder on-the-fly, called xmlns:xlink="http://www.w3.org/1999/xlink">adaptive perceptron coding , which continuously learns the quickly converges signal statistics. test adaptive over binary image database, results always exceeding performance LUT-based methods recurrent networks. also compare version requiring offline training, leads equally satisfactory results.