02/06/2020
The Human Resources Strategy for Researchers

PhD thesis, Learning societal preferences for automated collective decision making, M/W

This job offer has expired


  • ORGANISATION/COMPANY
    CNRS
  • RESEARCH FIELD
    Computer science
    MathematicsAlgorithms
  • RESEARCHER PROFILE
    First Stage Researcher (R1)
  • APPLICATION DEADLINE
    23/06/2020 23:59 - Europe/Brussels
  • LOCATION
    France › PARIS 16
  • TYPE OF CONTRACT
    Temporary
  • JOB STATUS
    Full-time
  • HOURS PER WEEK
    35
  • OFFER STARTING DATE
    01/09/2020

The PhD thesis will be done in the LAMSADE laboratory, in University Paris-Dauphine.

Collective decisions affect an important part of our lives. Some typical examples:

1. allocating scarce medical resources in sanitary crises (e.g., ventilators or hospital beds in the covid-19 pandemics, organ transplants).

2. deciding on how to spend the budget of some collectivity (nation, town, research group, group of co-owners), or part of it on different items or projects (e.g., participatory budgeting).

3. allocating students to schools or universities (e.g., in France, Affelnet ou Parcoursup).

4. autonomous devices or vehicles making decisions in situations of danger involving individuals' health or lives (e.g., dilemmas for autonomous cars)

5. implementing policies for incentivising people to behave in an environment-friendly way (e.g., large-scale vehicle-sharing, adaptive driving taxes that evolve with the level of pollution and the traffic)

Some of these decisions are automated (3,4, possibly 2 and 5), some can be semi-automated, in the sense that the decision is made by human experts or governing bodies after using computerised decision aiding techniques (typically 1), and some are taken in a classical, fully non-automated way (typically 2, 5).

All of these decisions involve some social choice or game-theoretic mechanisms: resource allocation (1), voting and portioning (2), matching with preferences (3), coalition formation (5), pricing mechanisms (5). Some involve uncertainty (often 1, 4). These mechanisms have as input the preferences or needs of concerned individuals (1, 2, 3), their characteristics (1, 3, 4), and/or the current state of the world (1, 4, 5).

A key question is whether the choice of the mechanism, or of its parameters, should be made centrally by governing bodies, by expert panels, by the society itself, or perhaps a mixture of these. The society can give their opinion or their preferences using classical polls or with crowdsourcing. Some existing work exist on crowdsourcing preferences for autonomous car dilemmas, organ matching or budgeting (Awad et al, 2020; Freedman et al, 2020; Garg et al. 2019) but there is no much more than that.

Some questions we want to address in the thesis are:

- give a taxonomy of collective decision problems where preferences or criteria can be crowdsourced, and formalise them in a social choice setting. (While this is simple for resource allocation or matching problems, this is less so for moral dilemmas, see below.)

- formalise moral dilemmas in a social choice and game theory setting.

- study the normative criteria or properties of the mechanisms for which it makes sense to elicit the opinion of the members of the society.

- how can we explain to individuals how a mechanisms works, and how well do people understand it?

- how can we learn the society's preferences on criteria or parameters of a social choice mechanism? How can we elicit numerical trade-offs between criteria? (a preliminary study is in Conitzer et al., 2016).

- learn societal preferences using sophisticated (e.g., non-additive) preference models.

- what can we do with the learned preferences? can we cluster users in groups of similar preferences? can we use this clustering into user types for saving time when eliciting preferences?

- design new mechanisms and algorithms that suit better societal preferences.

- investigate the strategic, game-theoretic issues of preference crowdsourcing.

The PhD thesis will consist both on theoretical work and implementation. It is interdisciplinary (economics, theoretical and applied computer science, AI) and will be concerned with the following research fields: economics (social choice theory, game theory, decision theory), AI (preference learning nd more generally machine learning, computational social choice) and classical computer science (both theoretic and applied).

References

Edmond Awad, Sohan Dsouza, Jean-François Bonnefon, Azim Shariff, Iyad Rahwan:
Crowdsourcing moral machines. Commun. ACM 63(3): 48-55 (2020)

Rachel Freedman, Jana Schaich Borg, Walter Sinnott-Armstrong, John P. Dickerson, Vincent Conitzer:
Adapting a kidney exchange algorithm to align with human values. Artif. Intell. 283: 103261 (2020)

Vincent Conitzer, Rupert Freeman, Markus Brill, Yuqian Li:
Rules for Choosing Societal Tradeoffs. AAAI 2016: 460-467

Nikhil Garg, Vijay Kamble, Ashish Goel, David Marn, Kamesh Munagala: Iterative Local Voting for Collective Decision-making in Continuous Spaces. J. Artif. Intell. Res. 64: 315-355 (2019)

Required Research Experiences

  • RESEARCH FIELD
    Computer science
  • YEARS OF RESEARCH EXPERIENCE
    None
  • RESEARCH FIELD
    MathematicsAlgorithms
  • YEARS OF RESEARCH EXPERIENCE
    None

Offer Requirements

  • REQUIRED EDUCATION LEVEL
    Computer science: Master Degree or equivalent
    Mathematics: Master Degree or equivalent
  • REQUIRED LANGUAGES
    FRENCH: Basic
Work location(s)
1 position(s) available at
Laboratoire d'Analyse et Modélisation de Systèmes pour l'Aide à la Décision
France
PARIS 16

EURAXESS offer ID: 528613
Posting organisation offer ID: 15586

Disclaimer:

The responsibility for the jobs published on this website, including the job description, lies entirely with the publishing institutions. The application is handled uniquely by the employer, who is also fully responsible for the recruitment and selection processes.

 

Please contact support@euraxess.org if you wish to download all jobs in XML.