Data demodulation using convolutional neural networks for holographic data storage

• Published in: Japanese Journal of Applied Physics Vol. 57; no. 9S1; pp. 9 - 13
• Main Authors: Katano, Yutaro, Muroi, Tetsuhiko, Kinoshita, Nobuhiro, Ishii, Norihiko, Hayashi, Naoto
• Format: Journal Article
• Language: English
• Published: Tokyo The Japan Society of Applied Physics 01.09.2018; Japanese Journal of Applied Physics
• Subjects: Artificial neural networks; Data storage; Data transfer (computers); Demodulation; Image classification; Machine learning; Neural networks; Noise tolerance; Optical noise; Optical properties
• ISSN: 0021-4922; 1347-4065
• DOI: 10.7567/JJAP.57.09SC01

We propose a data demodulation method based on a deep-learning algorithm. A convolutional neural network (CNN), which can accurately classify images, was used in the demodulation of data reproduced from holographic data storage (HDS). We designed CNNs and taught them the rules for demodulation based on the optical characteristics of the HDS using 700 reproduced data pages. The CNNs that learned could demodulate the data and decrease the number of demodulation errors by about 75% compared with hard decision image classification methods. This result showed an improvement in optical noise tolerance, which enhances the HDS with larger capacity and higher data-transfer rate.