Beginning Anomaly Detection Using

 Summary

In this chapter, we discussed practical use cases of anomaly detection in the business 

landscape. We showed how anomaly detection can be used to address real-life problems 

in many businesses. Every business and use case is different, so while we cannot copy/

Chapter 8   praCtiCal Use Cases of anomaly DeteCtion

---

318

paste code to build a successful model to detect anomalies in any dataset, this chapter 

covered many use cases to give you an idea of the possibilities and concepts behind the 

thought process.

Remember that this is an evolving field with continuous inventions and 

enhancements to the algorithms present, which means that in the future the 

algorithms will not look the same. Just couple of years ago, the RNN (recurrent neural 

network) was the best algorithm for a time series, but now the LSTM (Chapter 

6

) is 

being used heavily and the TCN (Chapter 

7

) will be the future of dealing with a time 

series. Even autoencoders have changed quite a bit; the traditional autoencoders have 

evolved into variational autoencoders (Chapter 

4

). The RBM (Chapter 

5

) is not used 

that much any longer.

In the next chapter, Appendix A, we will look at Keras, which is a popular framework 

for deep learning.

Chapter 8   praCtiCal Use Cases of anomaly DeteCtion

---

319

© 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