Redundancy Meets Synergy: Dependency-aware Expert Selection for MoE via Submodular Optimization

arXiv:2610.00558 · cs.LG, cs.DC · Submitted 2026-09-30 · Read on arXiv

cs.LG, cs.DC

Submitted: 2026-09-30

Updated: 2026-10-15

Comments: 26 pages, 3 figures

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• Redundancy — 1× in this paper · explained in 2 episodes

Sources

• NSC-SL: A Bandwidth-Aware Neural Subspace Compression for Communication-Efficient Split Learning
• FastMoE: A Fast Mixture-of-Expert Training System
• Mixtral of Experts
• Scaling Laws for Neural Language Models
• GAPSL: A Gradient-Aligned Parallel Split Learning over Data-Heterogeneous Edge Computing Systems
• DeepSeek-V3 Technical Report
• SEER-MoE: Sparse Expert Efficiency through Regularization for Mixture-of-Experts
• Optimizing Split Federated Learning with Unstable Client Participation
• ST-MoE: Designing Stable and Transferable Sparse Expert Models

Related papers

• Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
• AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
• Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
• Convergence issues in Relational Concept Analysis based on AOC-posets
• Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
• Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks