Learning Deep Learning: Theory and Practice of Neural Networks, Computer Vision, NLP, and Transformers using TensorFlow

Chapter 7

Convolutional Neural Networks Applied to Image Classification

Training of deep models with backpropagation has been demonstrated in various forms since at least 1990 (Hinton, Osindero, and Teh, 2006; Hinton and Salakhutdinov, 2006; LeCun et al., 1990; LeCun, Bottou, Bengio, et al., 1998). Still, a pivotal point for deep learning (DL) was in 2012 when AlexNet was published (Krizhevsky, Sutskever, and Hinton, 2012). It scored significantly better than any other contestant in the ImageNet classification challenge (Russakovsky et al., 2015) and greatly contributed to popularizing DL. AlexNet is an eight-layer network and uses convolutional layers, which were introduced by Fukushima (1980) and later used in LeNet (LeCun et al., 1990). Convolutional ...