ALPS: Adaptive Quantization of Deep Neural Networks with GeneraLized PositS

Hamed Fatemi Langroudi, Vedant Karia, Zachariah Carmichael, Abdullah M. Zyarah, Tej Pandit, John L. Gustafson, Dhireesha Kudithipudi

CVPRW 2021 pp. 3100-3109

doi:10.1109/CVPRW53098.2021.00346 /cvprw/2021/langroudi2021cvprw-alps/

Abstract

In this paper, a new adaptive quantization algorithm for generalized posit format is presented, to optimally represent the dynamic range and distribution of deep neural network parameters. Adaptation is achieved by minimizing the intra-layer posit quantization error with a compander. The efficacy of the proposed quantization algorithm is studied within a new low-precision framework, ALPS, on ResNet-50 and EfficientNet models for classification tasks. Results assert that the accuracy and energy dissipation of low-precision DNNs using generalized posits outperform other well-known numerical formats, including standard posits.

CVPRW Semantic Scholar

Cite

Text

Langroudi et al. "ALPS: Adaptive Quantization of Deep Neural Networks with GeneraLized PositS." IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, 2021. doi:10.1109/CVPRW53098.2021.00346

Markdown

[Langroudi et al. "ALPS: Adaptive Quantization of Deep Neural Networks with GeneraLized PositS." IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, 2021.](https://mlanthology.org/cvprw/2021/langroudi2021cvprw-alps/) doi:10.1109/CVPRW53098.2021.00346

BibTeX

@inproceedings{langroudi2021cvprw-alps,
  title     = {{ALPS: Adaptive Quantization of Deep Neural Networks with GeneraLized PositS}},
  author    = {Langroudi, Hamed Fatemi and Karia, Vedant and Carmichael, Zachariah and Zyarah, Abdullah M. and Pandit, Tej and Gustafson, John L. and Kudithipudi, Dhireesha},
  booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops},
  year      = {2021},
  pages     = {3100-3109},
  doi       = {10.1109/CVPRW53098.2021.00346},
  url       = {https://mlanthology.org/cvprw/2021/langroudi2021cvprw-alps/}
}