AI Paper Atlas

Research map / Multi-agent systems and games / Games and equilibria

Automated Dynamic Mechanism Design

Hanrui Zhang, Vincent Conitzer

NeurIPS 2021 · Multi-agent systems and games › Games and equilibria · approach: Learning with strategic agents · 4 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Account for the strategic agent's incentives instead of using the naïve policySynthetic environments with |S|=|A|=2 and η=-1 · Synthetic environments, |S|=|A|=2, η=-1principal utility (% of naïve benchmark): about 75% -> about 95%Improved
Account for the strategic agent's incentives instead of using the naïve policySynthetic environments with |S|=|A|=2 and η=0 · Synthetic environments, |S|=|A|=2, η=0principal utility (% of naïve benchmark): 85% -> about 95%Improved
Account for the strategic agent's incentives instead of using the naïve policySynthetic environments with larger state and action spaces and η=-1 · Synthetic environments, larger state and action spaces, η=-1principal utility (% of naïve benchmark): below 20% -> about 70%Improved
Account for the strategic agent's incentives instead of using the naïve policySynthetic environments with larger state and action spaces and η=0 · Synthetic environments, larger state and action spaces, η=0principal utility (% of naïve benchmark): about 50% -> about 70%Improved
Use a naïve mechanism optimized as if the agent reports truthfullySynthetic environments with larger state and action spaces · Synthetic dynamic environments, increasing state and action-space sizeprincipal utility (% of naïve benchmark): below 20%; about 50%Hurt
Use incentive-aware mechanisms as the environment grows more complexSynthetic dynamic environments · Synthetic environments, larger state and action spaces, η=-1principal utility (% of naïve benchmark): below 20% -> about 70%Improved
Use a naïve policy against a myopic agent rather than model its short-term incentivesSynthetic environments with aligned valuations (η=1) · Synthetic environments, η=1principal utilityHurt
Use a mechanism optimized for a myopic agentSynthetic environments with aligned valuations (η=1) · Synthetic environments, η=1, especially longer horizonsprincipal utilityImproved
Increase alignment between principal and agent valuationsNaïve and incentive-aware mechanisms · Synthetic environments, varying ηprincipal utilityImproved
Use an incentive-aware mechanism rather than a naïve mechanism as interests become misalignedSynthetic environments, varying η · Synthetic environments, varying valuation correlationprincipal utilityImproved
Increase the time horizonOptimal mechanisms facing strategic agents · Random synthetic dynamic environments, varying Tprincipal utilityNo clear effect
Face a myopic rather than patient agent when interests are fully adversarial (η=-1)Optimal mechanisms · Synthetic environments, η=-1principal utilityImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy