AI Paper Atlas

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

On Imitation in Mean-field Games

Giorgia Ramponi, Pavel Kolev, Olivier Pietquin, Niao He, Mathieu Lauriere, Matthieu Geist

NeurIPS 2023 · Multi-agent systems and games › Games and equilibria · approach: Mean-field game learning · 1 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
Replace vanilla adversarial IL (expert-population-driven) with MFC-ADV proxy that controls policy-induced occupancy measuresAdversarial imitation in MFGs · Attractor MFG, varying L and HNash imitation gap vs. proxy errorImproved
Increase population dependence of dynamics L (0.01 to 2) for BCBehavioral cloning · Attractor MFGNash imitation gap vs. BC errorHurt
Increase horizon H (3 to 100) for BCBehavioral cloning · Attractor MFGNash imitation gap vs. BC errorHurt
Increase L and H for vanilla-ADVVanilla adversarial IL · Attractor MFGNash imitation gap vs. vanilla-ADV errorHurt
Use vanilla-ADV vs MFC-ADV at small L and HAdversarial imitation · Attractor MFG, small L and HNash imitation gapNo clear effect
Use adversarial IL instead of BC when dynamics are population-independent (L_P=0)MFG imitation · Theory, L_P=0Nash imitation gap bound horizon dependenceImproved
Make transition dynamics population-dependent (L_P>0)BC and vanilla-ADV bounds · TheoryNash imitation gap bound horizon dependenceHurt

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