Presented By
O’Reilly + Cloudera
Make Data Work
29 April–2 May 2019
London, UK

Reinforcement learning: A gentle introduction and an industrial application

Christian Hidber (bSquare)
14:0514:45 Thursday, 2 May 2019
Data Science, Machine Learning & AI
Location: Capital Suite 15/16
Average rating: ****.
(4.86, 7 ratings)

Who is this presentation for?

  • Developers and data scientists



Prerequisite knowledge

  • Familiarity with machine learning and neural networks (useful but not required)

What you'll learn

  • Understand why reinforcement learning is a great complementary approach to supervised learning
  • Learn how this technique drastically improved an industrial application


Reinforcement learning (RL) learns complex processes autonomously like walking, beating the world champion in Go, or flying a helicopter. No big datasets with the “right” answers are needed: the algorithms learn by experimenting.

Christian Hidber shows how and why RL works and demonstrates how to apply it to an industrial hydraulics application with 7,000 clients in 42 countries—illustrating the inner workings by the way a child learns to play a new game. The industrial application stems from siphonic roof drainage systems, which ensure that large buildings like stadiums, airports, and shopping malls do not collapse during heavy rainfalls. Choosing the “right” diameters is difficult, requiring intuition and hydraulic expertise. As of today no feasible, deterministic algorithm is known. Christian explains how using reinforcement learning, he and his team were able to reduce the failure rate of the existing solution (based on classic supervised learning) by more than 70%.

Photo of Christian Hidber

Christian Hidber


Christian Hidber is a software engineer at bSquare, where he applies machine learning to industrial hydraulics simulation, part of a product with 7,000 installations in 42 countries. He holds a PhD in computer algebra from ETH Zurich, which he followed with a postdoc at UC Berkeley, where he researched online data mining algorithms.