GelGT proposes collaborative sampling and Gaussian attention on subgraphs to model long-range structural, semantic, and temporal dependencies in relational graphs, reporting up to 13.8% gains on downstream tasks.
Deep residual learning for image recognition
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
years
2026 4verdicts
UNVERDICTED 4representative citing papers
DIP-KD achieves state-of-the-art results in black-box data-free knowledge distillation across 12 benchmarks by synthesizing diverse image priors, applying contrastive learning, and using a primer student for soft-probability transfer.
LightSplit uses non-invertible orthogonal projections as an information bottleneck in split learning to reduce transmitted dimensionality by 32x while retaining more than 95% accuracy and limiting reconstruction risk.
An adaptive high-confidence image selection scheme during GAN training expands diversity in the distillation set for black-box few-shot KD and yields SOTA student accuracy on seven image datasets.
citing papers explorer
-
Gaussian Relational Graph Transformer
GelGT proposes collaborative sampling and Gaussian attention on subgraphs to model long-range structural, semantic, and temporal dependencies in relational graphs, reporting up to 13.8% gains on downstream tasks.
-
Diverse Image Priors for Black-box Data-free Knowledge Distillation
DIP-KD achieves state-of-the-art results in black-box data-free knowledge distillation across 12 benchmarks by synthesizing diverse image priors, applying contrastive learning, and using a primer student for soft-probability transfer.
-
LightSplit: Practical Privacy-Preserving Split Learning via Orthogonal Projections
LightSplit uses non-invertible orthogonal projections as an information bottleneck in split learning to reduce transmitted dimensionality by 32x while retaining more than 95% accuracy and limiting reconstruction risk.
-
Improving Diversity in Black-box Few-shot Knowledge Distillation
An adaptive high-confidence image selection scheme during GAN training expands diversity in the distillation set for black-box few-shot KD and yields SOTA student accuracy on seven image datasets.