[PlanetKR] Postdoc position on planning with others' beliefs and goals
Andreas HERZIG
andreas.herzig at irit.fr
Thu Apr 7 13:39:32 UTC 2022
* Job Description
Coming up with plans to achieve their goals---and more generally
deciding and learning how to act---are fundamental problems for
intelligent agents. The traditional focus of AI planning is the
generation of plans for an individual agent under the hypothesis that no
other agent intervenes and that the planning agent has perfect knowledge
of the environment and action pre- and postconditions. More expressive
forms of planning should include things such as reasoning about other
agents' beliefs, goals, intentions and actions ('theory of mind');
learning from past experiences; monitoring the actual outcome of actions
and learning possibly unexpected outcomes and dealing with such new
outcomes. Formal tools such as logics of knowledge and belief, logics of
goals, intentions and actions, game theory, and learning can be expected
to be building blocks of such formalisms.
Applicants should propose a short research project related to these
themes, possibly relating it to previous work of ours where we have
contributed to epistemic planning and to planning inspired from
Bratman's theory of intention.
The duration of the contract is 18 months (possibly with an extension).
Gross salary is between 2.663€ (about 2.131€ net) and 3.783€ (about
3.027€ net), depending on the experience of the candidate. Starting date
is September 1, 2022.
* Context
The Institut de Recherche en Informatique de Toulouse (IRIT) is one of
the biggest computer science labs in France. Its AI Department has a
long-standing tradition of research in knowledge representation and
reasoning. IRIT participates in the ANITI project on hybrid AI.
The postdoc position is within the TAILOR network (Trustworthy AI:
Integrating Learning, Optimization and Reasoning,
https://tailor-network.eu/) in which CNRS-IRIT takes part, and is part
of its workpackage WP5 "Deciding and Learning how to Act", possibly in
connection with WP6 "Learning and reasoning in social contexts". TAILOR
is one of four networks of excellence working on aspects of trustworthy
AI funded under the H2020-ICT-48-2020 call.
* Profile and Application
We are looking for a candidate with a background in knowledge
representation and/or machine learning who is interested in their
integration in the context of planning. Candidates should send the
following to Andreas.Herzig at irit.fr before April 30:
• CV;
• Up to 2 reference letters;
• A short proposal of research activities (max 1 page).
* Bibliography
• Michael Bratman: Intention, Plans, and Practical Reason. Harvard
University Press, Cambridge (1987). Reedited 1999 with CSLI Publications.
• Philip R. Cohen, Hector J. Levesque: Intention is choice with
commitment. J. Artif. Intell. 42(2), 213–261 (1990).
• Martin C. Cooper, Andreas Herzig, Faustine Maffre, Frédéric Maris,
Elise Perrotin, Pierre Régnier: A lightweight epistemic logic and its
application to planning. Artif. Intell. 298: 103437 (2021).
• Andreas Herzig, Emiliano Lorini, Laurent Perrussel, Zhanhao Xiao: BDI
Logics for BDI Architectures: Old Problems, New Perspectives. Künstliche
Intell. 31(1): 73-83 (2017).
More information about the PlanetKR
mailing list