Our paper "Heterogeneous Mean-Field Reinforcement Learning for Age-Minimal GPU Batching" is accepted by IEEE INFOCOM 2026

We are glad to share that our paper "Heterogeneous Mean-Field Reinforcement Learning for Age-Minimal GPU Batching" has been accepted to the IEEE International Conference on Computer Communications (INFOCOM 2026).

The paper introduces a heterogeneous mean-field reinforcement learning approach to GPU request batching that minimizes the age of information, scaling to many heterogeneous jobs that must be batched for efficient inference while keeping results fresh.