mlsanityData to 14 Oct 2026

DNN Memory Footprint Reduction via Post-Training Intra-Layer Multi-Precision Quantization

arXiv 2404.02947 ↗First posted 3 Apr 2024cs.LG cs.AIcited by 0

Not on the map. Why?

Abstract (arXiv)

The imperative to deploy Deep Neural Network (DNN) models on resource-constrained edge devices, spurred by privacy concerns, has become increasingly apparent. To facilitate the transition from cloud to edge computing, this paper introduces a technique that effectively reduces the memory footprint of DNNs, accommodating the limitations of resource-constrained edge devices while preserving model accuracy. Our proposed technique, named Post-Training Intra-Layer Multi-Precision Quantization (PTILMPQ), employs a post-training quantization approach, eliminating the need for extensive training data. By estimating the importance of layers and channels within the network, the proposed method enables precise bit allocation throughout the quantization process. Experimental results demonstrate that PTILMPQ offers a promising solution for deploying DNNs on edge devices with restricted memory resources. For instance, in the case of ResNet50, it achieves an accuracy of 74.57% with a memory footprint of 9.5 MB, representing a 25.49% reduction compared to previous similar methods, with only a minor 1.08% decrease in accuracy.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 16 of its 21 references that are in the map. First posted on arXiv 3 Apr 2024.

Builds on

16 of 21 references.

References · 16

  1. 2015Distilling the Knowledge in a Neural Networkcited by 10,324
  2. 2017MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applicationscited by 3,845
  3. 2017Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inferencecited by 1,593
  4. 2016Pruning Filters for Efficient ConvNetscited by 1,143
  5. 2016SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5MB model sizecited by 1,060
  6. 2020What is the State of Neural Network Pruning?cited by 583
  7. 2018Quantizing deep convolutional networks for efficient inference: A whitepapercited by 521
  8. 2019HAWQ: Hessian AWare Quantization of Neural Networks with Mixed-Precisioncited by 402
  9. 2019Data-Free Quantization Through Weight Equalization and Bias Correctioncited by 336
  10. 2017A Survey of Model Compression and Acceleration for Deep Neural Networkscited by 301
  11. 2020ZeroQ: A Novel Zero Shot Quantization Frameworkcited by 269
  12. 2018Faster gaze prediction with dense networks and Fisher pruningcited by 94
4 more
  1. 2020Post-Training Piecewise Linear Quantization for Deep Neural Networkscited by 79
  2. 2019Same, Same But Different - Recovering Neural Network Quantization Error Through Weight Factorizationcited by 45
  3. 2017Balanced Quantization: An Effective and Efficient Approach to Quantized Neural Networkscited by 21
  4. 2021Differentiable Dynamic Quantization with Mixed Precision and Adaptive Resolutioncited by 20

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2021ILMPQ : An Intra-Layer Multi-Precision Deep Neural Network Quantization framework for FPGA
  2. 2024Joint Pruning and Channel-wise Mixed-Precision Quantization for Efficient Deep Neural Networks
  3. 2022Edge Inference with Fully Differentiable Quantized Mixed Precision Neural Networks
  4. 2022Post-Training Quantization for Energy Efficient Realization of Deep Neural Networks
  5. 2023Performance Characterization of using Quantization for DNN Inference on Edge Devices: Extended Version
  6. 2026SigmaQuant: Hardware-Aware Heterogeneous Quantization Method for Edge DNN Inference
  7. 2021BMPQ: Bit-Gradient Sensitivity Driven Mixed-Precision Quantization of DNNs from Scratch
  8. 2025Sensitivity-Aware Post-Training Quantization for Deep Neural Networks