Applying machine learning to big data streams : An overview of challenges
Christoph Augenstein, Norman Spangenberg, Bogdan Franczyk · 2017
The importance of processing stream data increases with new technologies and new use cases. Applying machine learning to stream data and process them in real time leads to challenges in different ways. Model changes, concept drift or insufficient time to train models are a few examples. We illustrate big data characteristics and machine learning techniques derived from literature and conclude with available approaches and drawbacks.