Filters








440 Hits in 5.4 sec

Improving Lossless Compression Rates via Monte Carlo Bits-Back Coding [article]

Yangjun Ruan, Karen Ullrich, Daniel Severo, James Townsend, Ashish Khisti, Arnaud Doucet, Alireza Makhzani, Chris J. Maddison
2021 arXiv   pre-print
Latent variable models have been successfully applied in lossless compression with the bits-back coding algorithm.  ...  We demonstrate improved lossless compression rates in a variety of settings, especially in out-of-distribution or sequential data compression.  ...  Monte Carlo Bits-Back Coding The net bitrate of bits-back is ideally the negative ELBO. This rate seems difficult to improve without finding a better q.  ... 
arXiv:2102.11086v2 fatcat:ma5ywl4phrb6zfeuqq6jotvc4q

Error-diffused image compression using a binary-to-gray-scale decoder and predictive pruned tree-structured vector quantization

Ming Yuan Ting, E.A. Riskin
1994 IEEE Transactions on Image Processing  
This paper considers data compression of binary error diused images.  ...  The original contribution is using nonlinear lters to decode error diused images to compress them in the grayscale domain; this gives better image quality than directly compressing the binary images.  ...  We do this using empirical data in a Monte Carlo approach that is similar to standard VQ design.  ... 
doi:10.1109/83.336256 pmid:18296253 fatcat:ogfnrz2sobgv5pb5bdxp4onscu

Lossless Compression with Latent Variable Models [article]

James Townsend
2021 arXiv   pre-print
We develop a simple and elegant method for lossless compression using latent variable models, which we call 'bits back with asymmetric numeral systems' (BB-ANS).  ...  The method involves interleaving encode and decode steps, and achieves an optimal rate when compressing batches of data.  ...  Very recent work by Ruan et al. (2021) also directly extends the methods in this thesis, using Monte Carlo methods to improve the compression rate.  ... 
arXiv:2104.10544v2 fatcat:ndur24ecsbfxxjholb6aiakko4

An MCMC Approach to Lossy Compression of Continuous Sources

Dror Baron, Tsachy Weissman
2010 2010 Data Compression Conference  
Motivated by the Markov chain Monte Carlo (MCMC) relaxation method of Jalali and Weissman, we propose a lossy compression algorithm for continuous amplitude sources that relies on a finite reproduction  ...  Our algorithm asymptotically achieves the optimum rate distortion (RD) function universally for stationary ergodic continuous amplitude sources.  ...  In practice, many schemes rely on entropy coding, where scalar quantization is followed by lossless coding.  ... 
doi:10.1109/dcc.2010.11 dblp:conf/dcc/BaronW10 fatcat:cbg2gqghs5cspkxq2dmm74lwoa

Critical Data Compression [article]

John Scoville
2011 arXiv   pre-print
bits or compressing them using lossy methods.  ...  This method separates messages into components suitable for both lossless coding and 'lossy' or statistical coding techniques, compressing complex objects by separately encoding signals and noise.  ...  The original Monte Carlo problem, first solved by Metropolis and Hastings, dealt with the estimation of numerical integrals via sampling.  ... 
arXiv:1112.5493v1 fatcat:m5g6gm6qc5bjrk2uslhekcz46i

Split Hierarchical Variational Compression [article]

Tom Ryder, Chen Zhang, Ning Kang, Shifeng Zhang
2022 arXiv   pre-print
bits-back coding.  ...  This success, made possible by the bits-back coding framework, has produced competitive compression performance across many benchmarks.  ...  Monte-Carlo integration and the reparametrization trick [17] .  ... 
arXiv:2204.02071v1 fatcat:bavjymyqnnglro3pushiep6f54

On the Information Rates of the Plenoptic Function

Arthur L. da Cunha, Minh N. Do, Martin Vetterli
2010 IEEE Transactions on Information Theory  
We derive bounds on the lossless and lossy information rates for this dynamic reality model, stating conditions under which the bounds are tight.  ...  Examples with synthetic sources suggest that within our proposed model, common hybrid coding using motion/displacement estimation with DPCM performs considerably suboptimally relative to the true rate-distortion  ...  Here is inferred via Monte Carlo simulation with trials, and a minimum MSE detector for . The inferred is so low that the lower and upper bounds practically coincide.  ... 
doi:10.1109/tit.2009.2039168 fatcat:i66twsydengirbl37l3k6i465e

Data streams from the low frequency instrument on-board the PLANCK satellite: Statistical analysis and compression efficiency

M. Maris, D. Maino, C. Burigana, F. Pasian
2000 Astronomy and Astrophysics Supplement Series  
We find that such result is not improved by other lossless compressors, being the signal almost white noise dominated.  ...  by several lossless compression algorithms.  ...  Albrecht, for the useful suggestions and corrections, which improve significantly the text readability and accuracy.  ... 
doi:10.1051/aas:2000289 fatcat:phz7absxd5gozpwhxauso3dqai

ProIO: An Event-Based I/O Stream Format for Protobuf Messages [article]

D. Blyth, J. Alcaraz, S. Binet, S.V. Chekanov
2019 arXiv   pre-print
The performance of the ProIO concept for a dataset with Monte-Carlo event records used in high-energy physics was benchmarked and compared/contrasted with ROOT I/O.  ...  Various combinations of general-purpose compression and variable-length integer encoding available in protobuf were used to investigate the relationship between I/O performance and size-on-disk in a few  ...  The results show that the protobuf varint compression used in the ProIO files in fact does improve the rate of random access.  ... 
arXiv:1812.03967v2 fatcat:iqgxjnmyl5csbkjgxbmjofzkwy

Tracking performance in high multiplicities environment at ALICE [article]

David Rohr
2017 arXiv   pre-print
Such high data compression is impossible by lossless ZIP-like algorithms, but it must use results from online reconstruction, which in turn requires online calibration.  ...  In LHC Run 3, ALICE will increase the data taking rate significantly to 50\,kHz continuous read out of minimum bias Pb-Pb events.  ...  All GPU code is implemented in a generic way, such that the same code runs on the processor (possibly parallelized via OpenMP), on NVIDIA GPUs via CUDA, and on AMD GPUs via OpenCL [8] .  ... 
arXiv:1709.00618v1 fatcat:cu2thkzb55cipe4iu34rvy67cu

An MCMC Approach to Universal Lossy Compression of Analog Sources

Dror Baron, Tsachy Weissman
2012 IEEE Transactions on Signal Processing  
Motivated by the Markov chain Monte Carlo (MCMC) approach to the compression of discrete sources developed by Jalali and Weissman, we propose a lossy compression algorithm for analog sources that relies  ...  The latter algorithm is advantageous due to its reduced complexity and improved rates of convergence when employed on sources with a finite and small optimum reproduction alphabet.  ...  Entropy coding Many practical lossy compression algorithms employ entropy coding, where scalar quantization is followed by lossless compression (ECSQ).  ... 
doi:10.1109/tsp.2012.2206585 fatcat:t7mgssgy7beylkvbklfgzr57wy

cuSZ: An Efficient GPU-Based Error-Bounded Lossy Compression Framework for Scientific Data [article]

Jiannan Tian, Sheng Di, Kai Zhao, Cody Rivera, Megan Hickman Fulp, Robert Underwood, Sian Jin, Xin Liang, Jon Calhoun, Dingwen Tao, Franck Cappello
2020 arXiv   pre-print
It also improves the compression ratio by up to 3.48x on the tested data compared with another state-of-the-art GPU supported lossy compressor.  ...  Experiments show that our cuSZ improves SZ's compression throughput by up to 370.1x and 13.1x, respectively, over the production version running on single and multiple CPU cores, respectively, while getting  ...  In particular, we fine-tune the chunk size in Huffman coding and develop an adaptive method that selects 32-bit or 64-bit representation dynamically for Huffman code and can significantly improve the utilization  ... 
arXiv:2007.09625v1 fatcat:f4sq3abcmvehvm4b6wishmudie

Compressing Multisets with Large Alphabets [article]

Daniel Severo, James Townsend, Ashish Khisti, Alireza Makhzani, Karen Ullrich
2021 arXiv   pre-print
Compressing a multiset as an ordered sequence with off-the-shelf codecs is computationally more efficient, but has a sub-optimal compression rate, as bits are wasted encoding the order between symbols.  ...  Our method is compatible with any prefix-free code.  ...  This has been followed by a series of works proposing elaborations on the original idea for specific classes of latent variable models [13, 14, 15] and for improving the compression rate by using Monte  ... 
arXiv:2107.09202v1 fatcat:xpmxlyp2nfbkllsnrhcnbdvupa

Statistical Complexity Analysis of Turing Machine tapes with Fixed Algorithmic Complexity Using the Best-Order Markov Model

Jorge M. Silva, Eduardo Pinho, Sérgio Matos, Diogo Pratas
2020 Entropy  
We assessed the validity of our methodology on synthetic and real genomic data showing that it is tolerant to increasing rates of editions and block permutations.  ...  In this paper, we use a compression-based approach to measure global and local statistical complexity of specific Turing machine tapes with the same number of states and alphabet.  ...  Carlo 28,146 ± 20,348 0.22753 ± 0.42403 2 5 34,000,000 Monte Carlo 26,932 ± 20,092 0.22643 ± 0.42403 2 6 50,000,000 Monte Carlo 25,963 ± 19,856 0.22512 ± 0.42363 2 7 50,000,000 Monte Carlo 25,164  ... 
doi:10.3390/e22010105 pmid:33285880 fatcat:7iqeslhffja4ji2ef5j2t2lofm

Neural Communication Systems with Bandwidth-limited Channel [article]

Karen Ullrich, Fabio Viola, Danilo Jimenez Rezende
2020 arXiv   pre-print
First, instead of separating the sub-tasks of compression (source coding) and error correction (channel coding), we propose to model both jointly.  ...  Finally, sampling from the joint model is improved by introducing auxiliary latent variables in the decoder.  ...  Hinton & Van Camp (1993) show that via bits-back coding this code length (rate) can actually be achieved. This argument has further been hardened by Townsend et al.  ... 
arXiv:2003.13367v2 fatcat:xycgws3i7zhsvkipavy5boyraa
« Previous Showing results 1 — 15 out of 440 results