A method for the construction of minimum-redundancy codes

By D. Huffman. 1952

Subfields: Information Theory, Algorithms & Complexity

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

In Algorithms & Complexity

Citations in the corpus

Deep learning in neural networks: An overview (2014, background): “Adding another RNN to the stack improves a bound on the data’s description length – equivalent to the negative logarithm of its probability (Huffman, 1952; Shannon, 1948) – as long as there is remaining local learnable predictability in the data representation on the corresponding level of the…”
Fast Lossless Neural Compression with Integer-Only Discrete Flows (2022, background): “Based on the insight that the optimal code length for a single symbol x is − log p D ( x ) , many efﬁcient entropy codecs (Huffman, 1952; Duda, 2009; 2013) have been developed, and they have achieved nearly optimal code length given known data distribution.”
Differentiable Model Compression via Pseudo Quantization Noise (2021, methodology): “However, more efficient coding techniques exist when the entropy of the data is low, such as Huffman coding (Huffman 1952).”
On Learning to Think: Algorithmic Information Theory for Novel Combinations of Reinforcement Learning Controllers and Recurrent Neural World Models (2015, background): “real-valued prediction errors ei,τ = (predi(τ)− sensei(τ))(2) (in practical applications the errors will be given with limited precision), and encode each ei,τ by −logPe(ei,τ ) bits [108, 257].”
Community detection in graphs (2009, methodology): “Huffman coding (Huffman, 1952) is used to name vertices.”
Demystifying Parallel and Distributed Deep Learning (2018, methodology): “Deep Compression [91] also employs the lossless Huffman Coding [113] to further increase storage efficiency without impairing convergence.”
A Survey of Quantization Methods for Efficient Neural Network Inference (2021, neutral): “Huffman coding in particular is motivated by this [108].”
Self-Delimiting Neural Networks (2012, background): “Note that small size or low description length is equivalent to high probability, since the negative binary logarithm of the probability of some SLIM NN’s trace is essentially the number of bits needed to encode trace by Huffman coding [29].”
Synchronization and pattern separation in spiking associative memories and visual cortical areas (2004, methodology): “2 we have demonstrated that optimal matrix compression by Huffman or Golomb coding [74, 57] (see also appendices A.”
A programming language (1899, background): “y = (24, 60, 60) and if x = (1,2, 5 ) is the elapsed time in hours, minutes, and seconds, then _y_ _|_ x_ = 3725 is the elapsed time in seconds.”
Foundations of Signal Processing (2014, unknown): “Huffman codes were introduced in [47]; the book by Cover and Thomas [17] covers them in Section 5.”
Wavelets and their applications (2007, methodology): “This method, allied to Huffman type coding or semiarithmetic or arithmetic codings [HUF 52], [AHO 87], [MAL 98], [WIT 87], [HOW 92], leads to satisfactory compression performances, comparable with those of JPEG 92.”