Michael Schlechtinger defended his PhD Thesis

On Thursday, September 26th, Michael Schlechtinger has defended his PhD thesis on “Investigating, Predicting, and Mitigating Collusive Behavior
in Deep Reinforcement Learning-Based Pricing AIs”, supervised by Prof. Heiko Paulheim.

The thesis has been conducted in the KarekoKI project, an interdisciplinary project investigating the capabilities of pricing agents to learn collaborative strategies, and the implication on antitrust legislation. The project has been jointly led by Heiko Paulheim and Prof. Thomas Fetzer from the university's department of law.

In his thesis, Michael has developed and tested deep reinforcement learning based pricing agents, and shown that they are able to learn collaborative strategies, even when using their own price and revenue as signals, i.e., when being unable to observe and even unaware of other market participants. These findings challenge the applicability of today's antitrust laws. Moreover, he has investigated methods for collecting evidence on collective behavior from the observation of pricing agents.