Showing papers on "Canonical Huffman code published in 2020"

Source Coding for Text Transmission using a Deep Neural Network as a Lossy Compression Stage

Abstract: The continuous growth of traffic in the telecommunication networks has motivated the search for optimal source codes that can achieve high percentages of compression of the information to be transmitted. However, the compression rates are limited in the practice for the type of messages to encode. For this reason, new techniques have been developed in order to improve the compression rates of the traditional algorithms. In particular, source coding techniques based on computational intelligence algorithms are being studied lately. Hence, this paper proposes a new source coding technique for text compression based on two stages: the initial stage uses a deep neural network, called Text Embedding Neural Network, and the second stage uses a Canonical Huffman Code. The deep neural network increases the compression rate by controlling the level of syntax loss allowed in each message through a single adjustable parameter. This combination is able to reduce the size of the transmitted messages up to 30% with relation to only use traditional source coding algorithms.

Variable-Radix Coding of the Reals

Abstract: Recently proposed real number systems like Posits and Elias codes make use of tapered accuracy resulting from variable-length coding of exponents and significands. Several quite different interpretations of these number systems have been provided, though most often these rely on some combination of fixed- and variable-length codes for exponent and significand. We provide a new perspective on these number systems that unifies known representations while suggesting new ones. Our framework is based on multibit radix representations that encode the exponent in unary, the leading nonzero digit in a variablelength code, and the remaining digits in fixed-length binary code. We show how Posits, the various Elias codes, and ieee 754 like representations can be expressed in this framework. Moreover, we show that Posits and the Elias γ and δ codes represent the leading digit using the canonical Huffman code for a probability distribution given by Benford’s law, which governs the probability of leading digits. We further show that Posits correspond to the use of a fixed radix while Elias δ and ω codes are based on simple sequences of increasing radix. Our approach provides for an intuitive and uniform framework for representing numbers that reveals a visual mapping between codewords and the binary representation of real numbers obscured by prior frameworks. This new interpretation suggests a generalization of Posits and other number systems and provides simple rules for designing information-theoretically optimal codes.