AlexNet
One of the earliest works in popularizing the use of CNNs in large-scale image classification, AlexNet was proposed by Alex Krizhevsky and their co-authors in 2012. It was submitted as an entry to the ImageNet challenge in 2012 and significantly outperformed its runner-up with a 16% top-5 error rate. AlexNet consists of eight layers in the following order:
[INPUT -> CONV1 -> POOL1 -> CONV2 -> POOL2 -> CONV3 -> CONV5 -> CONV5 -> POOL3 -> FC6 -> FC7 -> FC8].
CONV1 is a convolutional layer with 96 filters of size 11 x 11. CONV2 has 256 filters of size 5 x 5, CONV3 and CONV4 have 384 filters of size 3 x 3, followed by CONV5 with 256 filters of size 3 x 3. All pooling layers, POOL1, POOL2, and POOL3, have 3 x 3 pooling filters. Both FC6 ...
Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.
Read now
Unlock full access