Introduction
In the first module of this learning path, Intro to Machine Learning with Keras, you learned how to create a neural network using the higher-level Keras API. In this module, we reimplement the model, training, evaluation, and prediction portions of the code from the first module, but this time we use lower-level TensorFlow concepts. Keep in mind that for many scenarios, Keras offers all the functionality you need. But if you find that you need more control than Keras offers, then understanding TensorFlow's lower level API gives you extra flexibility to customize your network and training routine, and to debug your code in more detail.
For this module, we assume that you're comfortable with Python, and that you completed the first module of this learning path or you're already familiar with Keras. We don't assume any knowledge of TensorFlow.
Let's get started!
Learning objectives
- Learn basic TensorFlow topics, such as tensors, variables, and automatic differentiation.
- Learn the difference between eager and graph execution.
- Reimplement the model and training loop of an existing Keras project using TensorFlow.
Prerequisites
- Knowledge of Python
- Basic knowledge about how to use Jupyter Notebooks
- Completion of module 1 of this learning path or knowledge of Keras