Publications
As the only research organization in the financial sector in several new technology areas, we are raising the status of
Hana Financial Group and achieving external achievements through world-renowned conferences.
Papers
Differentiable Neural Architecture Transformation For Reproducible Architecture Improvement
- Date of publication
- 2020.06.15
- Issuing agency (Year)
- arXiv(2020)
- Author
- Do-Guk Kim, Heung-Chang Lee
- Link
- https://arxiv.org/abs/2006.08231
Abstract
Recently, Neural Architecture Search (NAS) methods are introduced and show impressive performance on many benchmarks. Among those NAS studies, Neural Architecture Transformer (NAT) aims to improve the given neural architecture to have better performance while maintaining computational costs. However, NAT has limitations about a lack of reproducibility. In this paper, we propose differentiable neural architecture transformation that is reproducible and efficient. The proposed method shows stale performance on various architectures. Extensive reproducibility experiments on two datasets, i.e., CIFAR-10 and Tiny Imagenet, present that the proposed method definitely outperforms NAT and be applicable to other models and datasets
PDF View