Abstract
The TensorFlow library has come a long way from its first appearance. Especially in the last year, many more features have become available that can make the life of researchers a lot easier. Things like eager execution and Keras allow scientists to test and experiment much faster and debug models in ways that were not possible before. It is essential for any researcher to know those methods and know when it makes sense to use them. In this chapter, we will look at few of them: eager execution, GPU acceleration, Keras, how to freeze parts of a network and train only specific parts (used very often, especially in transfer learning and image recognition), and finally how to save and restore models already trained. Those technical skills will be very useful, not only to study this book, but in real-life research projects.
This is a preview of subscription content, log in via an institution.
Buying options
Tax calculation will be finalised at checkout
Purchases are for personal use only
Learn about institutional subscriptionsNotes
- 1.
https://www.tensorflow.org/guide/eager (accessed 17th January, 2019)
- 2.
You can find the notebook with the code in the book repository. To find it, go to the Apress book website and click on the Download Code button. The link points to the GitHub repository. The notebook is in the Chapter2 folder.
- 3.
You can find this article at https://goo.gl/hXKNnf to learn how to do it.
- 4.
The result was obtained when calling the function in a Google Colab notebook.
- 5.
The code has been inspired by the Google code in the Google Colab documentation.
- 6.
Check the official documentation for the example at https://keras.io/getting-started/faq/#how-can-i-freeze-keras-layers .
- 7.
- 8.
The example was inspired by the official Keras documentation at https://www.tensorflow.org/tutorials/keras/save_and_restore_models .
- 9.
Check the official documentation at https://goo.gl/SnKgyQ .
- 10.
Author information
Authors and Affiliations
Rights and permissions
Copyright information
© 2019 Umberto Michelucci
About this chapter
Cite this chapter
Michelucci, U. (2019). TensorFlow: Advanced Topics. In: Advanced Applied Deep Learning . Apress, Berkeley, CA. https://doi.org/10.1007/978-1-4842-4976-5_2
Download citation
DOI: https://doi.org/10.1007/978-1-4842-4976-5_2
Published:
Publisher Name: Apress, Berkeley, CA
Print ISBN: 978-1-4842-4975-8
Online ISBN: 978-1-4842-4976-5
eBook Packages: Professional and Applied ComputingApress Access BooksProfessional and Applied Computing (R0)