PhD pos.: Program induction methods for discovering decision strategies

The Rationality Enhancement Lab at the MPI for Intelligent Systems in
Tübingen is looking for a PhD student interested in developing program
induction methods for discovering optimal decision strategies. Our mission
is to reverse-engineer and enhance human intelligence. You will be part of
a larger project whose goals are to a) reverse-engineer the computational
mechanisms enabling people to discover and continuously refine their highly
efficient algorithms for planning, reasoning, and decision-making, b)
develop automatic methods for discovering optimal heuristics and general
principles of good decision-making, and to c) develop cognitive tutors that
teach people optimal cognitive strategies. Your primary focus will be to
develop and evaluate cognitively-inspired learning algorithms for
discovering effective decision strategies. The goal is to develop robust
program induction methods that can automatically discover interpretable
decision strategies that people can use to make better choices. You will
evaluate your methods by the performance and computational efficiency of
the discovered algorithms, its robustness to misspecification. Finally, we
will conduct training experiments to evaluate if teaching people the
discovered strategies enables them to make better decisions. For more
information, please see the attached job ad. If you are interested in this
position, please contact Dr. Falk Lieder (falk.lieder@tuebingen.mpg.de).

The PhD student (m/f) will receive a PhD funding contract equivalent in
remuneration to pay group E13, 65% of the Collective Wage Agreement for the
Public Service. An initial contract will be given for 3 years with
possibility of 1-year extension. The successful PhD candidate (m/f) should
have strong programming skills, and a solid background in computer science,
cognitive
science, or psychology, and previous experience with Bayesian machine
learning and/or reinforcement learning methods. Experience with some of the
following is a plus but not required: program induction, programming and
running online experiments, cognitive modeling, research on human
decision-making,probabilistic programming, and metareasoning.

Comments are closed.