Fall, 2022

• Efficient Tuning

Explore efficient methods for Visual Prompt Tuning (VPT).

• Accelerated Federated Learning

Accelerating the convergence of federeated learning.

• Adversarial Patch Defense

Find the common characteristics of adversarial patches.

• Efficient Adversarial Training

Accelerating adversarial training without losing robustness.

• Robust Training on Backdoored Data

Training clean models on backdoored data.

• Trojan Model Detection

How to detect if a model contains backdoor?

• Exploration on Unlearnable Examples

How to make data unlearnable?