Shiyu Chang

papers

1,945

total citations

papers (20)

The Lottery Tickets Hypothesis for Supervised and Self-Supervised Pre-Training in Computer Vision Models

CVPR 2021arXiv

136

citations

Uncovering the Disentanglement Capability in Text-to-Image Diffusion Models

CVPR 2023arXiv

122

citations

Decomposing Uncertainty for Large Language Models through Input Clarification Ensembling

ICML 2024arXiv

101

citations

PARP: Prune, Adjust and Re-Prune for Self-Supervised Speech Recognition

NEURIPS 2021arXiv

citations

Training Stronger Baselines for Learning to Optimize

NEURIPS 2020arXiv

citations

Harnessing the Spatial-Temporal Attention of Diffusion Models for High-Fidelity Text-to-Image Synthesis

ICCV 2023arXiv

citations

Understanding Interlocking Dynamics of Cooperative Rationalization

NEURIPS 2021arXiv

citations

Fairness Reprogramming

NEURIPS 2022arXiv

citations

Robust Mixture-of-Expert Training for Convolutional Neural Networks

ICCV 2023arXiv

citations

KVLink: Accelerating Large Language Models via Efficient KV Cache Reuse

NEURIPS 2025arXiv

citations

Quarantine: Sparsity Can Uncover the Trojan Attack Trigger for Free

CVPR 2022arXiv

citations

VSP: Diagnosing the Dual Challenges of Perception and Reasoning in Spatial Planning Tasks for MLLMs

ICCV 2025

citations

Selectivity Drives Productivity: Efficient Dataset Pruning for Enhanced Transfer Learning

NEURIPS 2023arXiv

citations

Correcting Diffusion Generation through Resampling

CVPR 2024arXiv

citations

Fictitious Synthetic Data Can Improve LLM Factuality via Prerequisite Learning

ICLR 2025arXiv

citations

Speech Self-Supervised Learning Using Diffusion Model Synthetic Data

ICML 2024

citations

Sparse Cocktail: Every Sparse Pattern Every Sparse Ratio All At Once

ICML 2024

citations

Shiyu Chang

papers (20)

TransGAN: Two Pure Transformers Can Make One Strong GAN, and That Can Scale Up

The Lottery Ticket Hypothesis for Pre-trained BERT Networks

Adversarial Robustness: From Self-Supervised Pre-Training to Fine-Tuning

The Lottery Tickets Hypothesis for Supervised and Self-Supervised Pre-Training in Computer Vision Models

Uncovering the Disentanglement Capability in Text-to-Image Diffusion Models

Decomposing Uncertainty for Large Language Models through Input Clarification Ensembling

PARP: Prune, Adjust and Re-Prune for Self-Supervised Speech Recognition

Training Stronger Baselines for Learning to Optimize

Harnessing the Spatial-Temporal Attention of Diffusion Models for High-Fidelity Text-to-Image Synthesis

Understanding Interlocking Dynamics of Cooperative Rationalization

Fairness Reprogramming

Robust Mixture-of-Expert Training for Convolutional Neural Networks

KVLink: Accelerating Large Language Models via Efficient KV Cache Reuse

Quarantine: Sparsity Can Uncover the Trojan Attack Trigger for Free

VSP: Diagnosing the Dual Challenges of Perception and Reasoning in Spatial Planning Tasks for MLLMs

Selectivity Drives Productivity: Efficient Dataset Pruning for Enhanced Transfer Learning

Correcting Diffusion Generation through Resampling

Fictitious Synthetic Data Can Improve LLM Factuality via Prerequisite Learning

Speech Self-Supervised Learning Using Diffusion Model Synthetic Data

Sparse Cocktail: Every Sparse Pattern Every Sparse Ratio All At Once

papers (20)

TransGAN: Two Pure Transformers Can Make One Strong GAN, and That Can Scale Up

The Lottery Ticket Hypothesis for Pre-trained BERT Networks

Adversarial Robustness: From Self-Supervised Pre-Training to Fine-Tuning

The Lottery Tickets Hypothesis for Supervised and Self-Supervised Pre-Training in Computer Vision Models

Uncovering the Disentanglement Capability in Text-to-Image Diffusion Models

Decomposing Uncertainty for Large Language Models through Input Clarification Ensembling

PARP: Prune, Adjust and Re-Prune for Self-Supervised Speech Recognition

Training Stronger Baselines for Learning to Optimize

Harnessing the Spatial-Temporal Attention of Diffusion Models for High-Fidelity Text-to-Image Synthesis

Understanding Interlocking Dynamics of Cooperative Rationalization

Fairness Reprogramming

Robust Mixture-of-Expert Training for Convolutional Neural Networks

KVLink: Accelerating Large Language Models via Efficient KV Cache Reuse

Quarantine: Sparsity Can Uncover the Trojan Attack Trigger for Free

VSP: Diagnosing the Dual Challenges of Perception and Reasoning in Spatial Planning Tasks for MLLMs

Selectivity Drives Productivity: Efficient Dataset Pruning for Enhanced Transfer Learning

Correcting Diffusion Generation through Resampling

Fictitious Synthetic Data Can Improve LLM Factuality via Prerequisite Learning

Speech Self-Supervised Learning Using Diffusion Model Synthetic Data

Sparse Cocktail: Every Sparse Pattern Every Sparse Ratio All At Once