AlexNet won the 2012 ImageNet competition by a record margin for image-recognition systems of its time.
AlexNet was created by Alex Krizhevsky, Ilya Sutskever, and Geoffrey Hinton. In the 2012 ImageNet Large Scale Visual Recognition Challenge, it achieved a top-five error rate of 15.3%, compared with 26.2% for the next-best entry. That difference drew widespread attention to deep convolutional neural networks.
The system used five convolutional layers followed by fully connected layers. It was trained on graphics-processing units, which made the large computation practical. AlexNet also used rectified linear units and dropout, techniques that helped improve training and generalization.
AlexNet did not invent convolutional neural networks; earlier systems such as LeNet-5 had already used them for handwritten digits. Its importance was demonstrating how deep networks, large datasets, and GPU computation could transform large-scale visual recognition.