Broad Learning not Deep Learning: Application of AI for the Prediction of Train Arrivals

The talk will present a solution that T-Systems has created for a large european railway operator to improve passenger information by predicting the arrival of trains in real time based on the trains’ current positions.

It will be shown how classical statistical machine learning approaches can be combined with artificial neural networks to solve the problem. Here, as in many other real world applications, “classical” neural networks are more applicable than approaches from Deep Learning.

Required audience experience

Basic Understanding of machine learning. Train travelling experience helpful.

Objective of the talk

The objective of the talk is to show how the data influences the choice of the best matching algorithms. This includes understanding the business domain, the size of the feature space, quality of data and non-functional requirements of the application.