Beginning Anomaly Detection Using

Figure A-41.  Finding the dot product of two placeholder variables c and d using 

the Keras back end

Figure A-42.  Finding the sum of c along different axes using the Keras back end

Figure A-43.  Finding the mean of c using the Keras back end

Those are just some of the most basic functions available through the back end. The 

complete list of backend functions is available at 

https://keras.io/backend/

.

Appendix A   intro to KerAs

---

360

 Summary

Keras is a great tool to help you easily get involved with creating, training, and testing deep 

learning models, and provides a great deal of functionality while abstracting away the 

complicated syntax that TensorFlow has. Keras by itself can be sufficient, but as the content 

gets more advanced, it's better to have the level of customization and flexibility that 

TensorFlow or PyTorch offers. Keras allows you to use a wide variety of functions through 

the back end, allowing you to write custom layers, custom models, metrics, loss functions, 

and so on, but for the most customization and flexibility in how you want your neural 

networks to be (especially if you want to make completely new types of neural networks), 

then either tf.keras + TensorFlow or PyTorch would be better suited for your needs.

Appendix A   intro to KerAs

---

361

© Sridhar Alla, Suman Kalyan Adari 2019 

S. Alla and S. K. Adari, Beginning Anomaly Detection Using Python-Based Deep Learning,  

https://doi.org/10.1007/978-1-4842-5177-5