Data to 14 Oct 2026

And the Bit Goes Down: Revisiting the Quantization of Neural Networks

arXiv 1907.05686 ↗First posted 12 Jul 2019cs.CVcited by 78

Not on the map. Why?

Abstract (arXiv)

In this paper, we address the problem of reducing the memory footprint of convolutional network architectures. We introduce a vector quantization method that aims at preserving the quality of the reconstruction of the network outputs rather than its weights. The principle of our approach is that it minimizes the loss reconstruction error for in-domain inputs. Our method only requires a set of unlabelled data at quantization time and allows for efficient inference on CPU by using byte-aligned codebooks to store the compressed weights. We validate our approach by quantizing a high performing ResNet-50 model to a memory size of 5MB (20x compression factor) while preserving a top-1 accuracy of 76.1% on ImageNet object classification and by compressing a Mask R-CNN with a 26x factor.

Key facts. Cited by 78 papers in the map (data through 14 Oct 2026). First posted on arXiv 12 Jul 2019.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2021: 26; 2022: 17; 2023: 13; 2024: 11; 2025: 9; 2026: 2.

78 papers.

Sort:

By area:

2021A Survey of Quantization Methods for Efficient Neural Network Inferencecited by 815
2021LeViT: a Vision Transformer in ConvNet's Clothing for Faster Inferencecited by 475
2021A White Paper on Neural Network Quantizationcited by 472
2021Pre-Trained Models: Past, Present and Futurecited by 319
2024GPTVQ: The Blessing of Dimensionality for LLM Quantizationcited by 82
2022A Comprehensive Survey on Model Quantization for Deep Neural Networks in Image Classificationcited by 65
2021VS-Quant: Per-vector Scaled Quantization for Accurate Low-Precision Neural Network Inferencecited by 64
2021ActNN: Reducing Training Memory Footprint via 2-Bit Activation Compressed Trainingcited by 61
2024VPTQ: Extreme Low-bit Vector Post-Training Quantization for Large Language Modelscited by 57
2021Implicit Neural Video Compressioncited by 52
2022NIRVANA: Neural Implicit Representations of Videos with Adaptive Networks and Autoregressive Patch-wise Modelingcited by 51
2021Differentiable Model Compression via Pseudo Quantization Noisecited by 45

38 more

2021Inspect, Understand, Overcome: A Survey of Practical Methods for AI Safetycited by 35
2021DKM: Differentiable K-Means Clustering Layer for Neural Network Compressioncited by 31
2021Knowledge Distillation as Semiparametric Inferencecited by 29
2022Neural Network Quantization with AI Model Efficiency Toolkit (AIMET)cited by 24
2023FineQuant: Unlocking Efficiency with Fine-Grained Weight-Only Quantization for LLMscited by 22
2021LegoDNN: Block-grained Scaling of Deep Neural Networks for Mobile Visioncited by 20
2023LUT-NN: Empower Efficient Neural Network Inference with Centroid Learning and Table Lookupcited by 20
2021RED : Looking for Redundancies for Data-Free Structured Compression of Deep Neural Networkscited by 19
2021RED++ : Data-Free Pruning of Deep Neural Networks via Input Splitting and Output Mergingcited by 19
2021LVAC: Learned Volumetric Attribute Compression for Point Clouds using Coordinate Based Networkscited by 19
2023AdaptiveNet: Post-deployment Neural Architecture Adaptation for Diverse Edge Environmentscited by 19
2021KDLSQ-BERT: A Quantized Bert Combining Knowledge Distillation with Learned Step Size Quantizationcited by 15
2021Knowledge Distillation Methods for Efficient Unsupervised Adaptation Across Multiple Domainscited by 15
2024Optimal Quantization for Matrix Multiplicationcited by 14
2024VQ4DiT: Efficient Post-Training Vector Quantization for Diffusion Transformerscited by 12
2023LifeLearner: Hardware-Aware Meta Continual Learning System for Embedded Computing Platformscited by 11
2022MKQ-BERT: Quantized BERT with 4-bits Weights and Activationscited by 10
2021Fixed-point Quantization of Convolutional Neural Networks for Quantized Inference on Embedded Platformscited by 7
2022LilNetX: Lightweight Networks with EXtreme Model Compression and Structured Sparsificationcited by 7
2021FAT: Learning Low-Bitwidth Parametric Representation via Frequency-Aware Transformationcited by 6
2022Restructurable Activation Networkscited by 6
2023PQA: Exploring the Potential of Product Quantization in DNN Hardware Accelerationcited by 6
2023Relationship between Model Compression and Adversarial Robustness: A Review of Current Evidencecited by 6
2025any4: Learned 4-bit Numeric Representation for LLMscited by 5
2022YONO: Modeling Multiple Heterogeneous Neural Networks on Microcontrollerscited by 4
2022Look-ups are not (yet) all you need for deep learning inferencecited by 4
2022Reconciling Security and Communication Efficiency in Federated Learningcited by 4
2022Hyperspherical Quantization: Toward Smaller and More Accurate Modelscited by 4
2023eDKM: An Efficient and Accurate Train-time Weight Clustering for Large Language Modelscited by 4
2022Quantized Sparse Weight Decomposition for Neural Network Compressioncited by 3
2023Compacting Binary Neural Networks by Sparse Kernel Selectioncited by 3
2025ViM-VQ: Efficient Post-Training Vector Quantization for Visual Mambacited by 3
2022Learning Low-Rank Representations for Model Compressioncited by 2
2023Self-Distilled Quantization: Achieving High Compression Rates in Transformer-Based Language Modelscited by 2
2024From Algorithm to Hardware: A Survey on Efficient and Safe Deployment of Deep Neural Networkscited by 2
2024An Empirical Investigation of Matrix Factorization Methods for Pre-trained Transformerscited by 2
2024MVQ:Towards Efficient DNN Compression and Acceleration with Masked Vector Quantizationcited by 2
2025Changing Base Without Losing Pace: A GPU-Efficient Alternative to MatMul in DNNscited by 2

50 most cited of 78 All 78 (JSON)