Recent Developments in Deep Learning Architecture from AlexNet to ResNet
September 27, 2016
The article on GitHub titled The 9 Deep Learning Papers You Need To Know About (Understanding CNNs Part 3) is not an article about the global media giant but rather the advancements in computer vision and convolutional neural networks (CNNs). The article frames its discussion around the ImageNet Large-Scale Recognition Challenges (ILSVRC), what it terms the “annual Olympics of computer vision…where teams compete to see who has the best computer vision model for tasks such as classification, localization, detection and more.” The article explains that the 2012 winners and their network (AlexNet) revolutionized the field.
This was the first time a model performed so well on a historically difficult ImageNet dataset. Utilizing techniques that are still used today, such as data augmentation and dropout, this paper really illustrated the benefits of CNNs and backed them up with record breaking performance in the competition.
In 2013, CNNs flooded in, and ZF Net was the winner with an error rate of 11.2% (down from AlexNet’s 15.4%.) Prior to AlexNet though, the lowest error rate was 26.2%. The article also discusses other progress in general network architecture including VGG Net, which emphasized depth and simplicity of CNNs necessary to hierarchical data representation, and GoogLeNet, which tossed the deep and simple rule out of the window and paved the way for future creative structuring using the Inception model.
Chelsea Kerwin, September 27, 2016
Sponsored by ArnoldIT.com, publisher of the CyberOSINT monograph
There is a Louisville, Kentucky Hidden Web/Dark Web meet up on September 27, 2016.
Information is at this link: https://www.meetup.com/Louisville-Hidden-Dark-Web-Meetup/events/233599645/