**Postdoc as Research Group Leader (m/f/d, E14 TV-L, 100%)**

Application deadline : 30.04.2021

The chair for “Foundations of Machine Learning Systems” (Robert C. Williamson) of the Department of Computer Sciences of the University of Tübingen invites applications for the position of a

to be filled as soon as possible. The position is paid according to E 14 TV-L (100%) and is limited to three years.

I am interested in machine learning from a broad systems perspective, and am looking for a postdoctoral scientist who is both interested in the mathematical foundations of machine learning systems, as well as how these mathematical foundations relate to the socio-technical systems that machine learning technology is becoming increasingly embedded into. The work is expected to require a blend of sophisticated mathematics as well as new conceptual ideas. Specific topics to be explored by the successful candidate include:

*Information*: Information theoretic limits of machine learning, from a geometric perspective, extending for example my work on information processing theorems, and the geometry of losses (see this earlier work too). The overarching goal of this work is to build a theory of machine learning problems (an analogy is that of the development of functional analysis in the 20th century relative to the 19th century notion of a function as a formula)*Data*: New theories of data - existing machine learning is largely built upon probability theory. But there are many reasons why this is not adequate. Richer theories require more sophisticated mathematics to handle situations (for example) where relative frequencies are not stable.*Society*: The relationship between the mathematical formalisms used to describe data, and hence what machine learning algorithms do, and societal constraints such as fairness (extending for example my work on Fairness Risk Measures). Consequently, determination of theoretical limits to fairness (building for example on my work on the cost of fairness) becomes important.*Context*: The conceptual / philosophical basis for machine learning systems; in particular, how can one represent the context in which data is gathered, and in which decisions or outputs are deployed? How can one reason about this, and how can one relate this to the mathematical formalisms implicit in the earlier bullet points? Further develop the ideas sketched in my HDSR commentary.

Candidates should hold a PhD in a suitable discipline, including computer science, mathematics, engineering, any quantitative science, or philosophy (if suitably quantitative).

Applications with the usual documents (motivation letter, CV, transcripts of records of all your degrees, your favorite publication) should be sent in electronic form (as a single PDF, at most 5 MB) to charlotte.wenner @uni-tuebingen.de by 30 April 2021. Applications should include the names of three referees who can comment on the candidate’s scientific work. Questions can be directed to bobwilliamsonoz @icloud.com.

The university seeks to raise the number of women in research and teaching and therefore urges qualified women academics to apply for these positions. Equally qualified applicants with disabilities will be given preference. The employment will be carried out by the central administration of the University of Tübingen.

In your application, please refer to Polytechnicpositions.com