Markus Kaiser

Dr. Markus Kaiser

Research Scientist

Siemens AG

University of Cambridge

Biography

My name is Markus Kaiser and I’m a Research Associate at the University of Cambridge and a Research Scientist at Siemens AG. I joined the group of Neil Lawrence, Carl Henrik Ek and Ferenc Huszár in the Computer Laboratory in Cambridge, UK, where we work on principled probabilistic machine learning and systems-design for real-world machine learning. As part of the learning systems group at Siemens AI, I put these thoughts into practice and apply machine learning in safety-critical industrial applications.

I’m excited about encoding expert knowledge into hierarchical probabilistic models to formulate informative prior assumptions. Informative priors facilitate inference and specify what models should learn from data, making them more data efficient and trustworthy. At Siemens, I have worked together with domain experts to create ML systems that are insightful for engineers and that can be relied on. I care about how machine learning software can be formulated to enable practitioners to easily construct probabilistic models and to embed them into existing systems. In my research, I explore how Bayesian non-parametric models can be composed to enforce abstract constraints, yield principled reasoning under uncertainty, and enable scalable and reliable inference.

Interests
  • Uncertainty propagation in hierarchical systems
  • Bayesian machine learning
  • (Deep) Gaussian processes
  • Trustworthy and interpretable ML
  • Software-design for ML libraries
Education
  • Doctorate in Computer Science, 2021

    Technical University of Munich

  • M.Sc. in Computer Science, 2017

    KTH Royal Institute of Technology

  • M.Sc. in Computer Science, 2016

    Technical University of Munich

  • B.Sc. in Computer Science, 2013

    Technical University of Munich

Experience

 
 
 
 
 
Research Associate
March 2021 – Present Cambridge, UK
As part of the AutoAI project, I’m developing new techniques for safely deploying, maintaining and explaining large scale machine learning systems. I care about aspects of system design, software engineering and ML modelling to achieve this goal. I am especially interested in how these tools can be used in the context of hierarchical Bayesian models and uncertainty propagation.
 
 
 
 
 
Research Scientist
September 2018 – Present Munich, Germany

I work on the application of my research and other state-of-the-art Bayesian models in industrial applications. My responsibilities include:

  • Bayesian modelling in interaction with domain experts
  • Good software-engineering practice in ML contexts
  • Internal knowledge transfer
  • Patent harvesting and collaboration on grant applications
 
 
 
 
 
PhD Candidate
September 2016 – November 2020 Munich, Germany

My research focused on encoding expert knowledge into hierarchical probabilistic models to facilitate inference and specify what to learn from data. I have worked on:

  • Hierarchical probabilistic models
  • Deep Gaussian processes
  • Reinforcement learning under uncertainty
  • Expert-interpretable models
 
 
 
 
 
Master’s Thesis
September 2015 – June 2016 Munich, Germany

Title: Incorporating Uncertainty into Reinforcement Learning through Gaussian Processes.

  • Model based reinforcement learning with Gaussian Processes
  • Propagation of predictive uncertainties
  • Evaluation on a bicycle benchmark

Projects

Bayesian Decomposition of Multi-Modal Dynamical Systems for Reinforcement Learning
In this extension, we demonstrate how semantic decompositions of dynamics models for Reinforcement Learning significantly increase data efficiency. We show how good model specification is critical for success and how the decomposition can be used for reward shaping.
Bayesian Decomposition of Multi-Modal Dynamical Systems for Reinforcement Learning