IC-POT replaces global rejection in partial OT with pointwise costs over both measures, admits a dual with local thresholds, reduces to balanced OT on augmented support, and shows gains in PU learning, domain adaptation, and satellite ocean data.
Sparsity-constrained optimal transport
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
fields
cs.LG 3verdicts
UNVERDICTED 3representative citing papers
Introduces Adaptive Clustering router for MoE models that scales features to identify tight expert clusters, yielding faster convergence, robustness to corruption, and performance gains.
Mixture-of-Control adaptively combines local and global control states in transformer fine-tuning by treating per-block states as experts in a sparse MoE setup to improve cross-block communication while keeping memory and compute costs comparable to prior state-based methods.
citing papers explorer
-
Take It or Leave It: Intent-Controlled Partial Optimal Transport
IC-POT replaces global rejection in partial OT with pointwise costs over both measures, admits a dual with local thresholds, reduces to balanced OT on augmented support, and shows gains in PU learning, domain adaptation, and satellite ocean data.
-
Tight Clusters Make Specialized Experts
Introduces Adaptive Clustering router for MoE models that scales features to identify tight expert clusters, yielding faster convergence, robustness to corruption, and performance gains.
-
Mixture-of-Control: State-Aware Fine-Tuning for Transformer-based Models
Mixture-of-Control adaptively combines local and global control states in transformer fine-tuning by treating per-block states as experts in a sparse MoE setup to improve cross-block communication while keeping memory and compute costs comparable to prior state-based methods.