mlsanityData to 7 Oct 2026

Fast Lossless Neural Compression with Integer-Only Discrete Flows

arXiv 2206.08869 ↗First posted 17 Jun 2022cs.LG cs.CV cs.ITcited by 4

Not on the map. Why?

Abstract (arXiv)

By applying entropy codecs with learned data distributions, neural compressors have significantly outperformed traditional codecs in terms of compression ratio. However, the high inference latency of neural networks hinders the deployment of neural compressors in practical applications. In this work, we propose Integer-only Discrete Flows (IODF), an efficient neural compressor with integer-only arithmetic. Our work is built upon integer discrete flows, which consists of invertible transformations between discrete random variables. We propose efficient invertible transformations with integer-only arithmetic based on 8-bit quantization. Our invertible transformation is equipped with learnable binary gates to remove redundant filters during inference. We deploy IODF with TensorRT on GPUs, achieving 10x inference speedup compared to the fastest existing neural compressors, while retaining the high compression rates on ImageNet32 and ImageNet64.

Key facts. Cited by 4 papers in the map (data through 7 Oct 2026). Builds on 41 of its 57 references that are in the map. First posted on arXiv 17 Jun 2022.

Builds on

41 of 57 references.

References · 41

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2020Denoising Diffusion Probabilistic Modelscited by 20,851
  3. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,160
  4. 2020Denoising Diffusion Implicit Modelscited by 9,655
  5. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,005
  6. 2016Densely Connected Convolutional Networkscited by 6,691
  7. 2018The Lottery Ticket Hypothesis: Finding Sparse, Trainable Neural Networkscited by 2,250
  8. 2013Estimating or Propagating Gradients Through Stochastic Neurons for Conditional Computationcited by 2,031
  9. 2016Density estimation using Real NVPcited by 2,029
  10. 2015Learning both Weights and Connections for Efficient Neural Networkscited by 1,998
  11. 2017Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inferencecited by 1,548
  12. 2018Glow: Generative Flow with Invertible 1x1 Convolutionscited by 1,539
29 more
  1. 2021Variational Diffusion Modelscited by 1,145
  2. 2016Pruning Filters for Efficient ConvNetscited by 1,129
  3. 2016XNOR-Net: ImageNet Classification Using Binary Convolutional Neural Networkscited by 829
  4. 2017Channel Pruning for Accelerating Very Deep Neural Networkscited by 697
  5. 2019Learned Step Size Quantizationcited by 604
  6. 2015BinaryConnect: Training Deep Neural Networks with binary weights during propagationscited by 571
  7. 2016DoReFa-Net: Training Low Bitwidth Convolutional Neural Networks with Low Bitwidth Gradientscited by 557
  8. 2016Learning Structured Sparsity in Deep Neural Networkscited by 538
  9. 2018PACT: Parameterized Clipping Activation for Quantized Neural Networkscited by 456
  10. 2017PixelCNN++: Improving the PixelCNN with Discretized Logistic Mixture Likelihood and Other Modificationscited by 415
  11. 2019HAWQ: Hessian AWare Quantization of Neural Networks with Mixed-Precisioncited by 387
  12. 2019Flow++: Improving Flow-Based Generative Models with Variational Dequantization and Architecture Designcited by 258
  13. 2019BIVA: A Very Deep Hierarchy of Latent Variables for Generative Modelingcited by 109
  14. 2020Bayesian Bits: Unifying Quantization and Pruningcited by 93
  15. 2019Practical Lossless Compression with Latent Variables using Bits Back Codingcited by 85
  16. 2019Integer Discrete Flows and Lossless Compressioncited by 82
  17. 2015Fast ConvNets Using Group-wise Brain Damagecited by 81
  18. 2020A Statistical Framework for Low-bitwidth Training of Deep Neural Networkscited by 51
  19. 2021Densely connected normalizing flowscited by 49
  20. 2020VFlow: More Expressive Generative Flows with Variational Data Augmentationcited by 38
  21. 2019HiLLoC: Lossless Image Compression with Hierarchical Latent Variable Modelscited by 36
  22. 2020IDF++: Analyzing and Improving Integer Discrete Flows for Lossless Compressioncited by 35
  23. 2019Compression with Flows via Local Bits-Back Codingcited by 34
  24. 2021Implicit Normalizing Flowscited by 28
  25. 2021iVPF: Numerical Invertible Volume Preserving Flow for Efficient Lossless Compressioncited by 27
  26. 2022Efficient-VDVAE: Less is morecited by 26
  27. 2021iFlow: Numerically Invertible Flows for Efficient Lossless Compression via a Uniform Codercited by 25
  28. 2020ResNet or DenseNet? Introducing Dense Shortcuts to ResNetcited by 12
  29. 2020Reducing the Computational Cost of Deep Generative Models with Binary Neural Networkscited by 6

Built on by

Cited by year

Citing papers per year: 2023: 4.

4 papers.

  1. 2023A Comprehensive Survey of Continual Learning: Theory, Method and Applicationcited by 860
  2. 2023Structured Pruning for Deep Convolutional Neural Networks: A surveycited by 122
  3. 2023Approximating Human-Like Few-shot Learning with GPT-based Compressioncited by 7
  4. 2023LAPP: Layer Adaptive Progressive Pruning for Compressing CNNs from Scratchcited by 2

Similar, not cited

Close in wording, no citation either way

  1. 2026NeuralLVC: Neural Lossless Video Compression via Masked Diffusion with Temporal Conditioning
  2. 2026ZipServ: Fast and Memory-Efficient LLM Inference with Hardware-Aware Lossless Compression
  3. 2024Unicorn: Unified Neural Image Compression with One Number Reconstruction
  4. 202570% Size, 100% Accuracy: Lossless LLM Compression for Efficient GPU Inference via Dynamic-Length Float (DFloat11)
  5. 2026Hybrid Compression: Integrating Pruning and Quantization for Optimized Neural Networks
  6. 2026PULSE: Unlocking Practical Image Compression on Single-Thread CPU
  7. 2026StateSMix: Online Lossless Compression via Mamba State Space Models and Sparse N-gram Context Mixing
  8. 2026SplitZip: Ultra Fast Lossless KV Compression for Disaggregated LLM Serving