Authors:
Shinji Niihara
1
;
2
and
Minoru Mori
1
Affiliations:
1
Faculty of Information Technology, Kanagawa Institute of Technology, Atsugi-shi, Kanagawa, Japan
;
2
SHARP Corporation, Sakai-shi, Osaka, Japan
Keyword(s):
Deep Neural Network, Image Recognition, Label, Tensor Representation, Adversarial Examples.
Abstract:
One-hot vectors representing correct/incorrect answer classes as {1/0} are usually used as labels for classification problems in Deep Neural Networks. On the other hand, a method using a tensor consisting of speech spectrograms of class names as labels has been proposed and reported to improve resistance to Adversarial Examples. However, effective representations for tensor-based labels have not been sufficiently studied. In this paper, we evaluate the effects of selections of image, complexity, and tensor size expansion on the tensor representation labels. Evaluation experiments using several databases and DNN models show that higher accuracies and tolerances can be achieved by improving tensor representations.