Fully Unsupervised Deep-Learning using Dimension Expansion for Nanophotonic Design
We introduce a fully unsupervised deep-learning method for inverse design of nanophotonics using dimension expansion to eliminate pre-generated datasets and post-processing. The effectiveness of the approach is validated on two distinct designs: Y-splitters and metalenses.
Authors
- Chia Wei Hsu (ORCID: https://orcid.org/0000-0002-9609-7155)
- Owen D. Miller (ORCID: https://orcid.org/0000-0003-2745-2392)
- Michelle L. Povinelli (ORCID: https://orcid.org/0009-0009-6570-6980)
- Lujia Zhong
- Shuo Huang (ORCID: https://orcid.org/0000-0001-6133-7027)
- Mahsa Torfeh (ORCID: https://orcid.org/0000-0002-0292-304X)
Publication Details
- Published
- 2025-01-01
- DOI
- https://doi.org/10.1364/cleo_fs.2025.ff131_6
- Primary Topic
- Photonic and Optical Devices
- Type
- article
- Field-Weighted Citation Impact
- 0.00