Download Chapter 4: WORKING WITH DATA
The deep-learning revolution is upon us, and it’s never been easier to use AI algorithms and neural networks for large-scale problem solving. You don’t need any experience in machine learning to get started; all that’s required is a standard computer, some popular open-source software tools, and this book.
Expert author Ron Kneusel starts you at the beginning with an introduction to Python, the language ubiquitous in machine learning. He then shows you how to create optimal datasets for training successful deep-learning models, before guiding you through more granular concepts like classic machine-learning algorithms, neural networks, and modern convolutional neural networks (CNNs). Practical knowledge builds on intuition as you work with standard toolkits to gain hands-on experience with various models, conduct experiments based on case studies and, ultimately, apply the book’s examples toward designing your own projects.
Along the way, you’ll learn:
- Fundamental DL concepts such as classes and labels, constructing datasets, and what a model does
- How to use standard open-source toolkits (sklearn and Keras/TensorFlow) along with standard datasets (e.g., MNIST and CIFAR-10)
- How to tune and evaluate the performance of a model via current standards of practice
- How to train models for classifying handwritten digits, images, audio sounds, and more
- How CNNs work, and how their outputs are affected by parameter choices
While no one book could possibly capture the entirety of this dynamic, ever-expanding field, Practical Deep Learning with Python will help you get your foot in the pool so you can feel the water – then give you the skills and confidence to dive all in.