This is the paper that started all of modern AI. When they say "This is the ImageNet moment for X," they mean this paper. I'm happy that Geoffrey Hinton allowed me to host his papers on ChapterPal. There are many of them that I would classify as must-reads. The series starts with "ImageNet Classification with Deep Convolutional Neural Networks" by Krizhevsky, Sutskever, and Hinton, 2012. The paper simply invented or used at scale for the first time what we just consider today as "How else would you do that?": - End-to-end neural network training for the task; - Large-scale pretraining on large unlabeled data and task-specific finetuning on a small labeled dataset; - ReLU activation; - Data augmentation; - Dropout; - "Local response normalization" that inspired later BatchNorm/LayerNorm layers. I remember when I read it in 2012, I got chills. chapterpal.com/s/e907a2a2/im…