Interpolating Convolutional Neural Networks Using Batch Normalization
Abstract
Perceiving a visual concept as a mixture of learned ones is natural for humans, aiding them to grasp new concepts and strengthening old ones. For all their power and recent success, deep convolutional networks do not have this ability. Inspired by recent work on universal representations for neural networks, we propose a simple emulation of this mechanism by purposing batch normalization layers to discriminate visual classes, and formulating a way to combine them to solve new tasks. We show that this can be applied for 2-way few-shot learning where we obtain between 4% and 17% better accuracy compared to straightforward full fine-tuning, and demonstrate that it can also be extended to the orthogonal application of style transfer.
Cite
Text
Wesley Putra Data et al. "Interpolating Convolutional Neural Networks Using Batch Normalization." Proceedings of the European Conference on Computer Vision (ECCV), 2018. doi:10.1007/978-3-030-01261-8_35Markdown
[Wesley Putra Data et al. "Interpolating Convolutional Neural Networks Using Batch Normalization." Proceedings of the European Conference on Computer Vision (ECCV), 2018.](https://mlanthology.org/eccv/2018/wesleyputradata2018eccv-interpolating/) doi:10.1007/978-3-030-01261-8_35BibTeX
@inproceedings{wesleyputradata2018eccv-interpolating,
title = {{Interpolating Convolutional Neural Networks Using Batch Normalization}},
author = {Wesley Putra Data, Gratianus and Ngu, Kirjon and William Murray, David and Adrian Prisacariu, Victor},
booktitle = {Proceedings of the European Conference on Computer Vision (ECCV)},
year = {2018},
doi = {10.1007/978-3-030-01261-8_35},
url = {https://mlanthology.org/eccv/2018/wesleyputradata2018eccv-interpolating/}
}