ReMP: Rectified Metric Propagation for Few-Shot Learning
Abstract
Few-shot learning features the capability of generalizing from a few examples. In this paper, we first identify that a discriminative feature space, namely a rectified metric space, that is learned to maintain the metric consistency from training to testing, is an essential component to the success of metric-based few-shot learning. Numerous analyses indicate that a simple modification of the objective can yield substantial performance gains. The resulting approach, called rectified metric propagation (ReMP), further optimizes an attentive prototype propagation network, and applies a repulsive force to make confident predictions. Extensive experiments demonstrate that the proposed ReMP is effective and efficient, and outperforms the state of the arts on various standard few-shot learning datasets.
Cite
Text
Zhao et al. "ReMP: Rectified Metric Propagation for Few-Shot Learning." IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, 2021. doi:10.1109/CVPRW53098.2021.00291Markdown
[Zhao et al. "ReMP: Rectified Metric Propagation for Few-Shot Learning." IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, 2021.](https://mlanthology.org/cvprw/2021/zhao2021cvprw-remp/) doi:10.1109/CVPRW53098.2021.00291BibTeX
@inproceedings{zhao2021cvprw-remp,
title = {{ReMP: Rectified Metric Propagation for Few-Shot Learning}},
author = {Zhao, Yang and Li, Chunyuan and Yu, Ping and Chen, Changyou},
booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops},
year = {2021},
pages = {2581-2590},
doi = {10.1109/CVPRW53098.2021.00291},
url = {https://mlanthology.org/cvprw/2021/zhao2021cvprw-remp/}
}