Abstract

This paper discusses an approach to the description of the structure of models capable of being trained to recognize representations of items of generative models—in particular, the architecture of a convolutional autoencoder is considered in detail. Reliable qualitative results of the operation of a convolutional encoder are also presented that show that it is valid to regard this model as generative because it is possible to implement output and sampling procedures, using as an example the solution of the problem of restoring images in missing regions.

© 2015 Optical Society of America

Autoencoder-based holographic image restoration

Tomoyoshi Shimobaba, Yutaka Endo, Ryuji Hirayama, Yuki Nagahama, Takayuki Takahashi, Takashi Nishitsuji, Takashi Kakue, Atsushi Shiraki, Naoki Takada, Nobuyuki Masuda, and Tomoyoshi Ito
Appl. Opt. 56(13) F27-F30 (2017)

Unsupervised segmentation of biomedical hyperspectral image data: tackling high dimensionality with convolutional autoencoders

Ciaran Bench, Jayakrupakar Nallala, Chun-Chin Wang, Hannah Sheridan, and Nicholas Stone
Biomed. Opt. Express 13(12) 6373-6388 (2022)

Photonic analog-to-digital converter powered by a generalized and robust convolutional recurrent autoencoder

Xiuting Zou, Shaofu Xu, Anyi Deng, Na Qian, Rui Wang, and Weiwen Zou
Opt. Express 28(26) 39618-39628 (2020)

Convolutional autoencoder for exposure effects equalization and noise mitigation in optical camera communication

Cristo Jurado-Verdu, Victor Guerra, Vicente Matus, Jose Rabadan, and Rafael Perez-Jimenez
Opt. Express 29(15) 22973-22991 (2021)

Processing of chromatic information in a deep convolutional neural network

Alban Flachot and Karl R. Gegenfurtner
J. Opt. Soc. Am. A 35(4) B334-B346 (2018)