AlexNet

Alex Krizhevsky, Ilya Sutskever & Geoffrey Hinton, 2012

AlexNet, created by Krizhevsky, Sutskever, and Hinton in 2012, dramatically won the ImageNet challenge and proved that deep convolutional networks trained on GPUs could vastly outperform hand-crafted features. The visualization shows its architecture as a sequence of layer blocks flowing left to right: five convolutional layers with pooling, followed by three fully connected layers. Block sizes reflect the spatial dimensions shrinking while channel depth grows. A gold wavefront animates data passing through each layer, settling to teal once processed.