Jun Zhuang
Boise State University, ID·Computer Science
Artificial intelligence models can fail in surprising ways. A small, deliberate change to an input can fool them, which is called an adversarial attack.
- Fields
Computer Science
Physics
- Topics
quantum neural networks
quantum machine learning
adversarial robustness
- Grants
- 1 known grant, through May 2027
What they work on
- Improve the resistance of quantum neural networks to adversarial attacks, meaning small deliberate input changes meant to fool them, using Bayesian label transition to infer more reliable labels.
- Propose new quantum models that handle open-set problems, where inputs fall outside the training categories, under perturbed or noisy conditions, and evaluate them on public datasets.
- Improve the stability of quantum neural network training by addressing gradient problems from an optimization perspective.
More about the research
They can also be confused by inputs unlike anything they saw in training, a situation called the open-set problem, and they can stall during training when the signals used to adjust them become unreliable. These weaknesses make AI less trustworthy in settings where errors matter. This project asks how to make AI models more robust. It focuses on quantum neural networks (QNNs), AI models that use the properties of quantum circuits to process data.
Why it matters
Many applications now depend on AI, including uses tied to national and social security. Models that can be fooled by tampered inputs, or that behave unpredictably on unfamiliar data, are a real risk in these settings. More robust quantum and classical neural networks could make future quantum AI systems more dependable for the people and organizations that rely on them.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Xinyu Xing
Northwestern University, IL·Computer Science·Early-career
Train adversarial agents that act in a target agent's environment to provoke undesired behavior, using explainable AI methods to guide where the attacks focus.
Sanjay Purushotham
University of Maryland-Baltimore County, MD·Computer Science·Early-career
Develop multi-view federated learning algorithms that train shared models across hospitals holding different kinds of patient data, without moving the data between sites.
Feng Chen
University of Texas at Dallas, TX·Computer Science
Develop online fairness-aware deep learning models for changing data streams, using causal representation learning, differentiable architecture search and online meta-learning, and release open software.
The school
Boise State University: 10 labs on PI Pages, $58M a year on research, 64 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.