Seunghoon Hong

papers

900

total citations

papers (19)

Pure Transformers are Powerful Graph Learners

NEURIPS 2022arXiv

258

citations

Part-Based Pseudo Label Refinement for Unsupervised Person Re-Identification

CVPR 2022arXiv

236

citations

SetVAE: Learning Hierarchical Composition for Generative Modeling of Set-Structured Data

CVPR 2021arXiv

104

citations

Improving Unsupervised Image Clustering With Robust Learning

CVPR 2021arXiv

citations

High-Fidelity Synthesis with Disentangled Representation

ECCV 2020arXiv

citations

Variational Interaction Information Maximization for Cross-domain Disentanglement

NEURIPS 2020arXiv

citations

Learning Probabilistic Symmetrization for Architecture Agnostic Equivariance

NEURIPS 2023arXiv

citations

Equivariant Hypergraph Neural Networks

ECCV 2022arXiv

citations

Learning to Compose: Improving Object Centric Learning by Injecting Compositionality

ICLR 2024arXiv

citations

Revisiting Random Walks for Learning on Graphs

ICLR 2025arXiv

citations

Towards End-to-End Generative Modeling of Long Videos With Memory-Efficient Bidirectional Transformers

CVPR 2023arXiv

citations

Chameleon: A Data-Efficient Generalist for Dense Visual Prediction in the Wild

ECCV 2024arXiv

citations

Transformers meet Stochastic Block Models: Attention with Data-Adaptive Sparsity and Cost

NEURIPS 2022arXiv

citations

MetaWeather: Few-Shot Weather-Degraded Image Restoration

ECCV 2024arXiv

citations

3D Denoisers Are Good 2D Teachers: Molecular Pretraining via Denoising and Cross-Modal Distillation

AAAI 2025arXiv

citations

Seunghoon Hong

papers (19)

Pure Transformers are Powerful Graph Learners

Part-Based Pseudo Label Refinement for Unsupervised Person Re-Identification

SetVAE: Learning Hierarchical Composition for Generative Modeling of Set-Structured Data

Improving Unsupervised Image Clustering With Robust Learning

High-Fidelity Synthesis with Disentangled Representation

Variational Interaction Information Maximization for Cross-domain Disentanglement

Learning Probabilistic Symmetrization for Architecture Agnostic Equivariance

Equivariant Hypergraph Neural Networks

Learning to Compose: Improving Object Centric Learning by Injecting Compositionality

Revisiting Random Walks for Learning on Graphs

Towards End-to-End Generative Modeling of Long Videos With Memory-Efficient Bidirectional Transformers

Chameleon: A Data-Efficient Generalist for Dense Visual Prediction in the Wild

Transformers meet Stochastic Block Models: Attention with Data-Adaptive Sparsity and Cost

MetaWeather: Few-Shot Weather-Degraded Image Restoration

3D Denoisers Are Good 2D Teachers: Molecular Pretraining via Denoising and Cross-Modal Distillation

Multi-View Representation Learning via Total Correlation Objective

Bridging the gap to real-world language-grounded visual concept learning

Transformers Generalize DeepSets and Can be Extended to Graphs & Hypergraphs

Disentangled Representation Learning via Modular Compositional Bias

papers (19)

Pure Transformers are Powerful Graph Learners

Part-Based Pseudo Label Refinement for Unsupervised Person Re-Identification

SetVAE: Learning Hierarchical Composition for Generative Modeling of Set-Structured Data

Improving Unsupervised Image Clustering With Robust Learning

High-Fidelity Synthesis with Disentangled Representation

Variational Interaction Information Maximization for Cross-domain Disentanglement

Learning Probabilistic Symmetrization for Architecture Agnostic Equivariance

Equivariant Hypergraph Neural Networks

Learning to Compose: Improving Object Centric Learning by Injecting Compositionality

Revisiting Random Walks for Learning on Graphs

Towards End-to-End Generative Modeling of Long Videos With Memory-Efficient Bidirectional Transformers

Chameleon: A Data-Efficient Generalist for Dense Visual Prediction in the Wild

Transformers meet Stochastic Block Models: Attention with Data-Adaptive Sparsity and Cost

MetaWeather: Few-Shot Weather-Degraded Image Restoration

3D Denoisers Are Good 2D Teachers: Molecular Pretraining via Denoising and Cross-Modal Distillation

Multi-View Representation Learning via Total Correlation Objective

Bridging the gap to real-world language-grounded visual concept learning

Transformers Generalize DeepSets and Can be Extended to Graphs & Hypergraphs

Disentangled Representation Learning via Modular Compositional Bias