AI & Computing · 2012
ImageNet Classification with Deep Convolutional Neural Networks (AlexNet)PDF Download
Krizhevsky, Sutskever, Hinton
AlexNet shows how a deep convolutional network, trained with GPUs and practical regularization choices, can perform strongly on large-scale image classification. Architecture and engineering matter together here.
The paper is a useful reminder that a breakthrough may be a combination of ingredients. Look at training details and error analysis alongside the headline benchmark result.
1.4 MB · SHA-256: 90137160c57217953d5f61857e64ca58e85f06e1b13b4f475c918b1b582b9771