On the Claim for the Existence of “Adversarial Examples” in Deep Learning Neural Networks

Costas Neocleous, Christos N. Schizas

2014

Abstract

A recent article in which it is claimed that adversarial examples exist in deep artificial neural networks (ANN) is critically examined. The newly discovered properties of ANNs are critically evaluated. Specifically, we point that adversarial examples can be serious problems in critical applications of pattern recognition. Also, they may stall the further development of artificial neural networks. We challenge the absolute existence of these examples, as this has not been universally proven yet. We also suggest that ANN structures, that correctly recognize adversarial examples, can be developed.

References

  1. http://www.image-net.org/ Seen in June 2014.
  2. http://www.nytimes.com/2014/05/31/us/on-death-rowwith-low-iq-and-new-hope-for-a-reprieve.html?_r=0 Seen in June 2014.
  3. LeCun Y., Cortes C., Burges C., The MNIST database of handwritten digits. http://yann.lecun.com/exdb/mnist/ Seen in June 2014.
  4. Marois R., Chun M., Gore J., 2000, Neural Correlates of the Attentional Blink, Neuron, Vol. 28, pp. 299-308.
  5. Minsky M., Papert S., 1972, Perceptrons: An Introduction to Computational Geometry, (2nd edition with corrections, first edition 1969), The MIT Press, Cambridge MA.
  6. Neokleous K., Avraamides M., Neocleous C., Schizas C., 2009. A neural network model of the attentional blink phenomenon. International Journal of Engineering Intelligent Systems for Electrical Engineering and Communications, Vol. 17, pp. 115-126.
  7. Szegedy C., Zaremba W., Sutskever I., Bruna J., Erhan D., Goodfellow I., Fergus R., 2014. Intriguing properties of neural networks, 2nd International Conference on Learning Representations, Rimrock Resort, Canada, April 14-16, 2014.
Download


Paper Citation


in Harvard Style

Neocleous C. and N. Schizas C. (2014). On the Claim for the Existence of “Adversarial Examples” in Deep Learning Neural Networks . In Proceedings of the International Conference on Neural Computation Theory and Applications - Volume 1: NCTA, (IJCCI 2014) ISBN 978-989-758-054-3, pages 306-309. DOI: 10.5220/0005152503060309


in Bibtex Style

@conference{ncta14,
author={Costas Neocleous and Christos N. Schizas},
title={On the Claim for the Existence of “Adversarial Examples” in Deep Learning Neural Networks},
booktitle={Proceedings of the International Conference on Neural Computation Theory and Applications - Volume 1: NCTA, (IJCCI 2014)},
year={2014},
pages={306-309},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0005152503060309},
isbn={978-989-758-054-3},
}


in EndNote Style

TY - CONF
JO - Proceedings of the International Conference on Neural Computation Theory and Applications - Volume 1: NCTA, (IJCCI 2014)
TI - On the Claim for the Existence of “Adversarial Examples” in Deep Learning Neural Networks
SN - 978-989-758-054-3
AU - Neocleous C.
AU - N. Schizas C.
PY - 2014
SP - 306
EP - 309
DO - 10.5220/0005152503060309