Lei Yu

Exploring backdoor attacks and graph reduction methods in GNNs.

Introducing DiscoGP, a new method for better understanding language models.

A new benchmark project aims to assess Java issue resolution capabilities.

A new method enhances diffusion models for better image quality and speed.

New training method improves LLM safety and performance.

A new method improves vulnerability detection in smart contracts.

Learn how to protect GNNs from adversarial attacks and enhance their reliability.

Discover how new methods improve code review comments through automation and evaluation.