Learning Deep Learning: Theory and Practice of Neural Networks, Computer Vision, Nlp, and Transformers Using Tensorflow

Ekman begins by introducing the perceptron and other artificial neurons, the fundamental building blocks of the deep neural networks that have enabled the deep learning revolution. He introduces fully connected feedforward networks and convolutional networks, showing how to apply them to solve practical problems, such as predicting housing prices or classifying images. He demonstrates how to represent words from a natural language using an encoding that captures semantics, and how to combine these together with a recurrent neural network to create a neural based natural language translator that can automatically translate simple French sentences to English. Then, building on all you've learned, Ekman guides you through building an image captioning network capable of inputting an image and generating a natural language description of it.

Throughout, you'll find concise, well-annotated code examples using TensorFlow and the Keras API; for comparison and easy migration between frameworks, complementary examples in PyTorch are provided online. Ekman also explains enough of the mathematics to help newcomers grasp how deep learning actually works. The guide concludes by previewing emerging trends in deep learning, and exploring the challenging ethical issues surrounding its use.