ML takes control – Converting Business Rules into Machine Learning Features

The ongoing trend towards Machine Learning (ML) based systems challenges existing decision engines which typically rely on a set of human-defined rules. Once there is enough historical data available for training, ML-based systems usually outperform human-defined rule sets easily.

In practice, however, there are many problems with the shift towards such systems. Typical questions are: When do I have enough training data for a certain feature to switch a rule off? How can I train a model if there was a manual rule that biased my ML training data?

This talk gives an overview of these problems and presents solutions that can be taken to move quickly towards ML-based decision engines.

Vorkenntnisse

* Grundlegendes Verständnis von Machine Learning.

Lernziele

* Die Teilnehmer sollen verstehen, wie sie ihre bisherigen Rule-based-Algorithmen in ML-Algorithmen transformieren können.