Dissecting Convolutional Neural Networks

ECU Author/Contributor (non-ECU co-authors, if there are any, appear on document)

Justin Daniel Whitaker (Creator)

Institution

East Carolina University (ECU )

Web Site: http://www.ecu.edu/lib/

Abstract: I examined the hidden layers of a convolutional neural network for MNIST handwritten digit classification. I found that it is possible to view how backpropagation affects nodes in the activation maps of hidden layers by visualizing their outputs as images across training epochs. This can allow one to look inside the black box of neural networks and to gain a deeper understanding of their mechanics.

Additional Information

Publication

Thesis

Language: English

Date: 2019

Keywords

machine learning, neural network, convolutional, supervised learning

Subjects

Email this document to