Xiaojun Jia

This paper examines the risk of backdoor attacks in few-shot learning.

Squeezed light improves the speed and security of quantum random number generators.

New method strengthens privacy for shared images and text.

Introducing TA-Cleaner, a method to improve multimodal model defenses against data poisoning.

Exploring innovative methods to enhance multispectral object detection accuracy.

Learn how external knowledge improves language model accuracy and reliability.