Yifei Wang

Dissecting the Diffusion Process in Linear Graph Convolutional Networks

2021 · arXiv (Cornell University)

Graph Convolutional Networks (GCNs) have attracted more and more attentions in recent years. A typical GCN layer consists of a linear feature propagation step and a nonlinear transformation step. Recent works show that a linear …

Nationwide Adoption of a Digital Contact Tracing App: Examining the Role of Privacy Concern, Political Trust, and Technology Literacy

2022 · Communication Studies

Digital contact tracing (DCT) has become an efficient method for authorities to curb COVID-19. However, the current public opinion studies on the adoption of DCT primarily focus on Western countries and rarely explore citizen attitudes …

Efficient semi-quantum secret sharing protocol using single particles

2023 · Chinese Physics B

Semi-quantum secret sharing (SQSS) is a branch of quantum cryptography which only requires the dealer to have quantum capabilities, reducing the difficulty of protocol implementation. However, the efficiency of the SQSS protocol still needs to …

Robust Long-Tailed Learning via Label-Aware Bounded CVaR

2023 · arXiv (Cornell University)

Data in the real-world classification problems are always imbalanced or long-tailed, wherein the majority classes have the most of the samples that dominate the model training. In such setting, the naive model tends to have …

OODRobustBench: a Benchmark and Large-Scale Analysis of Adversarial Robustness under Distribution Shift

2023 · arXiv (Cornell University)

Existing works have made great progress in improving adversarial robustness, but typically test their method only on data from the same distribution as the training data, i.e. in-distribution (ID) testing. As a result, it is …

An Expectation-Maximization Algorithm for Training Clean Diffusion Models from Corrupted Observations

2024 · arXiv (Cornell University)

Diffusion models excel in solving imaging inverse problems due to their ability to model complex image priors. However, their reliance on large, clean datasets for training limits their practical use where clean data is scarce. …

DyKen-Hyena: Dynamic Kernel Generation via Cross-Modal Attention for Multimodal Intent Recognition

2025 · arXiv (Cornell University)

Though Multimodal Intent Recognition (MIR) proves effective by utilizing rich information from multiple sources (e.g., language, video, and audio), the potential for intent-irrelevant and conflicting information across modalities may hinder performance from being further improved. …