Abstract

We studied the possibility of extracting object hierarchy information from a trained neural network by analyzing the errors obtained on a test sample using an approach based on singular value decomposition of the confusion matrix. Experiments indicate that the methods investigated in this paper can be used to obtain a tentative clustering of classes. In addition, we show that the number of connections within a fully connected layer of a convolutional neural network can be reduced without adversely affecting recognition accuracy for a test sample using locally connected layers. At the same time, however, our experiments did not show that a layer organization consistent with the object hierarchy led to any improvement of the results.

© 2017 Optical Society of America

Intensity and phase stacked analysis of a Φ-OTDR system using deep transfer learning and recurrent neural networks

Ceyhun Efe Kayan, Kivilcim Yuksel Aldogan, and Abdurrahman Gumus
Appl. Opt. 62(7) 1753-1764 (2023)

Infrared object classification with a hybrid optical convolution neural network

Jianbo Chen, Jennifer Talley, and Kevin F. Kelly
Appl. Opt. 60(25) G224-G231 (2021)

Learning spectral initialization for phase retrieval via deep neural networks

David Morales, Andrés Jerez, and Henry Arguello
Appl. Opt. 61(9) F25-F33 (2022)

Deep learning-based object classification through multimode fiber via a CNN-architecture SpeckleNet

Ping Wang and Jianglei Di
Appl. Opt. 57(28) 8258-8263 (2018)

Ghost imaging of blurred object based on deep-learning

Zijin Zhang, Chunfang Wang, Wenlin Gong, and Dawei Zhang
Appl. Opt. 60(13) 3732-3739 (2021)