Learn about Keras as a front-end to deep learning tools.
- [Narrator] Neural networks and deep learning are behind many of the recent breakthroughs in areas like image recognition, language translation, and speech recognition. Keras is a framework for building deep neural networks with Python. With Keras, you can build state-of-the-art, deep learning systems just like those used at Google and Facebook. Keras is designed to make it as easy as possible to build deep learning systems with as little complexity as possible. With Keras, you can build a deep neural network with only with only a few lines of code.
Keras doesn't do all of the work itself. It's really a front-end layer written in Python that runs on top of other popular deep learning toolkits like TensorFlow and Theano. At abstracts away a lot of the complexity of using those tools while still giving you many of the benefits. When you tell Keras to build a deep neural network, behind the scenes it builds out the neural network using either TensorFlow or Theano. In this course, we'll be using TensorFlow as the back-end. When you use Keras with TensorFlow, it builds a TensorFlow model and runs the training process for you.
That means that your model is compatible with most tools and utilities that work with TensorFlow. You can even upload your Keras model to Google's Cloud Machine learning system. One of the core principles of Keras is that best practices are built in. When building a deep learning system, there are many different parameters you have to configure. Keras always tries to provide good defaults for parameters. The default setting used in Keras are based on what has worked well for researchers in the past.
So more often than not, using the default settings in Keras will get you close to your goal. Even better, Keras comes with several pre-trained deep learning models for image recognition. You can use the pre-trained models to recognize common types of objects and images, or you can adapt these models to create a custom image recognition system with your own data.
- What's Keras?
- Using Keras vs. TensorFlow
- Training a deep learning model
- Using a pre-trained deep learning model
- Monitoring a Keras model with TensorBoard
- Using a trained Keras model in Google Cloud