ImageNet Classification with Deep Convolutional Neural Networks .

Abstract

AlexNet sparked the deep learning revolution by winning ImageNet 2012 by a massive margin. It demonstrated that deep CNNs could outperform traditional computer vision methods.

Architecture

• 5 convolutional layers
• 3 fully connected layers
• ReLU activation (first major use)
• Dropout for regularization
• Data augmentation

Impact

This paper made deep learning mainstream in computer vision and triggered the AI boom of the 2010s.