AI Paper Atlas

Research map / Reinforcement learning / Policy gradient and actor-critic

Constrained Multi-Objective Reinforcement Learning with Max-Min Criterion

Giseung Park, Hyunyoung Nam, Woohyeon Byeon, Amir Leshem, Youngchul Sung

ICML 2026 · Reinforcement learning › Policy gradient and actor-critic · approach: Model-based RL with guarantees · 0 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
add the w update (max-min weight learning) (ablation)Constrained max-min MORL · Building thermal controlminimum return: 626.7 -> 639.8Improved
add the u update (constraint multiplier learning) (ablation)Constrained max-min MORL · Building thermal controlcost sum: 200.7 -> 178.7Improved
add both u and w updates (ablation)Constrained max-min MORL · Building thermal controlcost sum: 222.0 -> 178.7Improved
add u update (ablation)Constrained max-min MORL · Building thermal controlminimum return: 653.5 -> 639.8Hurt
add constraint handling to max-minConstrained max-min vs unconstrained max-min · Tabular constrained MOMDPsoptimal value error: 0.325 -> 0.004Improved
add max-min fairness to constrained max-averageConstrained max-min vs constrained max-average · Tabular constrained MOMDPsoptimal value error: 0.657 -> 0.004Improved
reduce beta from 0.03 to 0.001 (ablation)Constrained max-min · Tabular constrained MOMDPsoptimal value error: 0.004 -> 0.021Hurt
increase beta from 0.03 to 0.1 (ablation)Constrained max-min · Tabular constrained MOMDPsoptimal value error: 0.004 -> 0.061Hurt
use ours instead of MA-SAC-LConstrained max-min vs MA-SAC-L · Building thermal controlminimum return: 620.9 -> 639.8Improved
use ours instead of MA-SAC-LConstrained max-min vs MA-SAC-L · MoAnt-v5minimum return: 83.0 -> 92.2Improved
use ours instead of MA-CPGOConstrained max-min vs MA-CPGO · Traffic signal controlminimum return: -27,830 -> -25,229Improved

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