Tag
The paper proposes 'unit' as an explicit primitive in machine learning, where learning tasks declare persistent individuals, and supervised learning specializes to unit-conditioned response laws with tokenization.
Proposes FedTCR, the first systematic federated multimodal graph learning algorithm that handles task, modality, and topology heterogeneity via topology-aware cross-modal routing and tri-level contrastive learning, outperforming baselines across 7 domains.
Proposes a Collaborative Meta Knowledge Enhancement (COME) framework for dementia etiology diagnosis that injects heterogeneity-aware embeddings into a unified Transformer architecture, achieving state-of-the-art performance across multiple independent cohorts.
Introduces HERO, a heterogeneity-aware benchmark library for federated continual learning that separates task splits, client data splits, and client task sequences to enable reproducible and setting-aware evaluation.
This paper analyzes the effect of structural and temporal heterogeneities in decentralized federated learning over temporal networks, showing that ignoring these heterogeneities leads to unrealistically rapid convergence and that real-world networks slow down diffusion.
This paper reanalyzes a prior study claiming lower AI literacy predicts greater AI receptivity, finding that the aggregate negative relationship masks heterogeneity: the effect is not significant for text AI tools but remains strong for non-text AI tools, indicating a narrower pattern of broader adoption rather than general receptivity.