Book Name: Applied Deep Learning
Author: Umberto Michelucci
File size: 12.5 MB
File format: PDF
Applied Deep Learning Pdf Book Description:
Work with complex subjects in profound learning, for example optimization calculations, hyper-parameter tuning, dropout, and error analysis in addition to approaches to deal with common issues encountered when coaching neural networks that are deep. You will start by analyzing the activation functions mainly with one neuron (ReLu, sigmoid, and Swish), viewing the way to carry out linear and logistic regression with TensorFlow, and deciding on the proper price function. The following section discusses more complex neural network architectures with various layers and nerves and investigates the issue of arbitrary initialization of weights.
Applied Deep Learning also discusses how to execute logistic regression entirely from scratch with no Python library NumPy, to allow you to enjoy how libraries like TensorFlow allow rapid and effective experiments. Case studies for each strategy are added to put to practice each of theoretical info. You will discover tips and techniques for writing optimized Python code (such as vectorizing loops together with NumPy).