AlexNet is a convolutional neural network architecture proposed by Alex Krizhevsky, Ilya Sutskever, and Geoffrey Hinton in 2012.

At the time, it achieved state-of-the-art performance on the test set for the 2010 ImageNet Large Scale Visual Recognition Competition (LSVRC). A variant of the model won the 2012 ImageNet LSVRC with a top-5 test error rate of 15.3%–ten percentage points ahead of the second place winner.