Deep Learning Hash for Wireless Multimedia Image Content Security

Table 1

A huge number of parameters need to be learned and stored with different datasets. Considering a classification task with C different categories and D-dimensional feature representation, even the simplest linear models are comprised of D×C parameters.

Image Dataset	Categories	Dimensions	Parameters
ImageNet	21,841	4,096	89,460,736
ILSVRC	1,000	4,096	4,096,000
SUN397	397	12,288	4,878,336
CIFAR-10	10	4,096	40,960
Caltech-256	256	4,096	1,048,576