The SNU ECE Optical Engineering and Quantum Electronics (OEQE) Laboratory Laboratory 3D Display Research Team (Advisor: Professor Yoonchan Jeong) suggested the possibility of manifold learning in the hologram domain. The researchers used an existing autoencoder structure to compress and play complex holograms. Unlike previous studies, their strategy was to train the neural network in the hologram domain instead of the image domain. The latent vector representation of a hologram corresponds to well-separated manifolds, showing the similarities between numbers successfully. This pipeline provides the foundation for research on generative models, such as the variational autoencoder, for enhancing manifold learning. This study was presented at the International Conference of 3D Systems and Applications (Part of The 30th International Display Workshops) (December 8th, 2023 / held in Niigata Prefecture, Japan), and was awarded the Best Poster Award.

[Presented Paper]

Lee, Y. Jeong, B. Lee, "Manifold learning in the hologram domain: a pipeline for compressing and reconstructing complex holograms", in International Conference of 3D Systems and Applications 2023 (Part of International Display Workshops (IDW) 2023), (3DSAp2-3), 3DSA/IDW, December 2023.

(left) Researcher Juhyun Lee of Optical Engineering and Quantum Electronics Laboratory,

(right) Research on pipeline for manifold learning in the hologram domain

Source:  https://ece.snu.ac.kr/ece/news?md=v&bbsidx=54245

Translated by: Jiyong Yoo, English Editor of the Department of Electrical and Computer Engineering, cyoo7@snu.ac.kr