Given nonnegative weights (wr, . . . , w,,), the well-known algorithm of Huffman [2] can be used to construct a binary tree with n external nodes and n 1 internal nodes, where the external nodes are labeled with the weights (wr, . . . , w,) in some order. Huffman’s tree has the minimum value of WJ, + * * . + w,,l, over all such binary trees, where lj is the level at which wj occurs in the tree....

Deep Compression is a three stage compression pipeline: pruning, quantization and Huffman coding. Pruning reduces the number of weights by 10x, quantization further improves the compression rate between 27x and 31x. Huffman coding gives more compression: between 35x and 49x. The compression rate already included the metadata for sparse representation. Deep Compression doesn’t incur loss of accu...

1 1 Abstract—This paper proposes a novel model of the two-level scalar quantizer with extended Huffman coding. It is designed for the average bit rate to approach the source entropy as close as possible provided that the signal to quantization noise ratio (SQNR) value does not decrease more than 1 dB from the optimal SQNR value. Assuming the asymmetry of representation levels for the symmetric ...

We complete the classification of all Lee-extremal and Lee-optimal self-dual codes over F2 + uF2 of lengths 9 through 20 that have a nontrivial odd order automorphism begun in [W.C. Huffman, On the decomposition of self-dual codes over F2 + uF2 with an automorphism of odd prime order, Finite Fields Appl., in press]. Along the way, we find all Lee-extremal self-dual codes over F2 + uF2 of length...

Upper bounds on the redundancy of Huffman codes have been extensively studied in the literature. Almost all of these bounds are in terms of the probability of either the most likely or the least likely source symbol. In this correspondence, we prove a simple upper bound in terms of the probability of any source symbol.