Books to Read and Tools

Books to Read

We believe that you learn best when you are exposed to multiple perspectives on the same idea. As such, we recommend checking out a few of the books below to get an added perspective on Deep Learning.

Grokking Deep Learning by Andrew Trask. Use our exclusive discount code traskud17 for 40% off. This provides a very gentle introduction to Deep Learning and covers the intuition more than the theory.

Neural Networks And Deep Learning by Michael Nielsen. This book is more rigorous than Grokking Deep Learning and includes a lot of fun, interactive visualizations to play with.

Inside Deep Learning by Edward Raff is an excellent intermediate book that covers much of the mathematical background, in-depth looks at modern neural network architectures, and develops a good intuition.

The Deep Learning Textbook from Ian Goodfellow, Yoshua Bengio, and Aaron Courville. Often simply called "the Deep Learning book", it is a rigorous treatment of the mathematics and theory behind deep learning, in addition to covering a number of important practical issues.

Deep Learning Architectures by Ovidiu Calin is by far the most theoretical book on this list, and the most challenging in terms of mathematical prerequisites. However, if you want to conduct groundbreaking research in neural networks, this book is an excellent reference to have on hand.

Deep Learning Tools

In this video, we covered a number of tools:

Development tools

• Integrated Development Environment
  • Code Editor
  • Interpreter/Compiler
• Jupyter Notebooks
  • Note: Each cell is executed on its own, and it works well for environments to prototype or present code. However, there are limitations:
    • Editing can make you lose state
    • Code deployed production should be in .py rather than notebooks

Deep Learning Frameworks

• PyTorch (aka Torch)
• TensorFlow/Keras
• JAX

Training Tools

• Experiment management like TensorBoard or Weights and Biases
  • Observe accuracy and loss at training time
• Model versioning like DVC, Neptune, and Pachyderm
  • Remedy issues within the model across different versions of the model
  • DVC is very similar to Git