Learning How Much to Think: Difficulty-Aware Dynamic MoEs for Graph Node Classification

ArXi:2604.11473v1 Announce Type: new Mixture-of-Experts (MoE) architectures offer a scalable path for Graph Neural Networks (GNNs) in node classification tasks but typically rely on static and rigid routing strategies that enforce a uniform expert budget or coarse-grained expert toggles on all nodes. This limitation overlooks the varying discriminative difficulty of nodes and leads to under-fitting for hard nodes and redundant computation for easy ones.