# Precision Gating: Improving Neural Network Efficiency with Dynamic Dual-Precision Activations

arXiv: https://arxiv.org/abs/2002.07136 · HTML: https://mlsanity.com/paper/2002.07136/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 17 Feb 2020.

## Built on by (most cited first, 5 of 5)

- [Prompt-Driven Dynamic Object-Centric Learning for Single Domain Generalization](https://mlsanity.com/paper/2402.18447/) (2024)
- [Enabling Design Methodologies and Future Trends for Edge AI: Specialization and Co-design](https://mlsanity.com/paper/2103.15750/) (2021)
- [Mathematical Formalism for Memory Compression in Selective State Space Models](https://mlsanity.com/paper/2410.03158/) (2024)
- [RCT: Resource Constrained Training for Edge AI](https://mlsanity.com/paper/2103.14493/) (2021)
- [Radial Networks: Dynamic Layer Routing for High-Performance Large Language Models](https://mlsanity.com/paper/2404.04900/) (2024)

## Abstract (arXiv)

We propose precision gating (PG), an end-to-end trainable dynamic dual-precision quantization technique for deep neural networks. PG computes most features in a low precision and only a small proportion of important features in a higher precision to preserve accuracy. The proposed approach is applicable to a variety of DNN architectures and significantly reduces the computational cost of DNN execution with almost no accuracy loss. Our experiments indicate that PG achieves excellent results on CNNs, including statically compressed mobile-friendly networks such as ShuffleNet. Compared to the state-of-the-art prediction-based quantization schemes, PG achieves the same or higher accuracy with 2.4× less compute on ImageNet. PG furthermore applies to RNNs. Compared to 8-bit uniform quantization, PG obtains a 1.2% improvement in perplexity per word with 2.7× computational cost reduction on LSTM on the Penn Tree Bank dataset. Code is available at: https://github.com/cornell-zhang/dnn-gating