# Focused Quantization for Sparse CNNs

arXiv: https://arxiv.org/abs/1903.03046 · HTML: https://mlsanity.com/paper/1903.03046/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 7 Mar 2019.

## Built on by (most cited first, 6 of 6)

- [Training Deep Neural Networks with Joint Quantization and Pruning of Weights and Activations](https://mlsanity.com/paper/2110.08271/) (2021)
- [Dynamic Probabilistic Pruning: A general framework for hardware-constrained pruning at different granularities](https://mlsanity.com/paper/2105.12686/) (2021)
- [MinUn: Accurate ML Inference on Microcontrollers](https://mlsanity.com/paper/2210.16556/) (2022)
- [FireFly-S: Exploiting Dual-Side Sparsity for Spiking Neural Networks Acceleration with Reconfigurable Spatial Architecture](https://mlsanity.com/paper/2408.15578/) (2024)
- [Rapid Model Architecture Adaption for Meta-Learning](https://mlsanity.com/paper/2109.04925/) (2021)
- [Deep Lookup Network](https://mlsanity.com/paper/2509.13662/) (2025)

## Abstract (arXiv)

Deep convolutional neural networks (CNNs) are powerful tools for a wide range of vision tasks, but the enormous amount of memory and compute resources required by CNNs pose a challenge in deploying them on constrained devices. Existing compression techniques, while excelling at reducing model sizes, struggle to be computationally friendly. In this paper, we attend to the statistical properties of sparse CNNs and present focused quantization, a novel quantization strategy based on power-of-two values, which exploits the weight distributions after fine-grained pruning. The proposed method dynamically discovers the most effective numerical representation for weights in layers with varying sparsities, significantly reducing model sizes. Multiplications in quantized CNNs are replaced with much cheaper bit-shift operations for efficient inference. Coupled with lossless encoding, we built a compression pipeline that provides CNNs with high compression ratios (CR), low computation cost and minimal loss in accuracy. In ResNet-50, we achieved a 18.08x CR with only 0.24% loss in top-5 accuracy, outperforming existing compression methods. We fully compressed a ResNet-18 and found that it is not only higher in CR and top-5 accuracy, but also more hardware efficient as it requires fewer logic gates to implement when compared to other state-of-the-art quantization methods assuming the same throughput.