The Emergence of Semantics in Neural Network Representations of Visual Information
Full Text: N18-2122.pdf
Word vector models learn about semantics through corpora. Convolutional Neural Networks (CNNs) can learn about semantics through images. At the most abstract level, some of the information in these models must be shared, as they model the same real-world phenomena. Here we employ techniques previously used to detect semantic representations in the human brain to detect semantic representations in CNNs. We show the accumulation of semantic information in the layers of the CNN, and discover that, for misclassified images, the correct class can be recovered in intermediate layers of a CNN.
Citation
D. Dharmaretnam, A. Fyshe. "The Emergence of Semantics in Neural Network Representations of Visual Information". NAACL Annual Conference of the North American Chapter of the Association for Computational Linguisti, New Orleans, USA, pp 776–780, June 2018.| Keywords: | |
| Category: | In Conference |
| Web Links: | DOI: |
| ACL |
BibTeX
@incollection{Dharmaretnam+Fyshe:NAACL18,
author = {Dhanush Dharmaretnam and Alona Fyshe},
title = {The Emergence of Semantics in Neural Network Representations of
Visual Information},
Pages = {776–780},
booktitle = {NAACL Annual Conference of the North American Chapter of the
Association for Computational Linguisti},
year = 2018,
}Last Updated: June 22, 2020Submitted by Sabina P
