RESEARCH
QUANTUM MACHINE LEARNING
Quantum machine learning (QML) has emerged as a promising paradigm for consequential machine learning tasks like image classification and generation. QML leverages the principles of superposition and entanglement inherent to quantum systems to explore vast data spaces and process them as complex quantum states. Some QML applications are even suitable for near-term quantum computing technology.
However, it is challenging to design and train models for QML tasks to run on small-to-medium-scale and high-noise quantum computers. Our research has focused on designing techniques for image classification and generation to run on current quantum computers while mitigating the effects of these challenges.
OPTIC [DATE 2022], QUILT [AAAI 2022], SLIQ [AAAI 2023], MOSAIQ [ICCV 2023]
NEUTRAL ATOM QUANTUM COMPUTING
Most of the focus in quantum compilation has been on superconducting qubits due to the ease of porting classical semiconductor technology to quantum computing. The current quantum software stack is designed to minimize the effect of hardware noise on current quantum computers by optimizing for these properties, and it is not suitable for other promising quantum computing technologies.
One such technology we've worked on is neutral atom quantum computing. Neutral atom qubits have some desirable properties that must be considered for optimizing quantum software compilation and stack solutions: multi-qubit. gates, customizable and flexible qubit connections, and long coherence times. Our research has focused on leveraging these properties to synthesize smaller gates into larger gates and construct application-specific qubit topologies.
GEYSER [ISCA 2022], GRAPHINE [SC 2023]
QUANTUM PRIVACY AND SECURITY
In the near future, only a few entities in the world may have access to powerful quantum computers, and these quantum computers will be used to solve previously-unsolved problems of significance, possibly without an explicit trust model between the service provider and the customer. However, the quantum code and programs that are shared and executed over the quantum cloud are not immune to adversarial snooping and attacks.
We anticipate that the solutions to such large-scale optimization problems will be considered sensitive and will need to be protected. Our work takes the first few steps toward preparing us for that future – by developing a novel method that intelligently obfuscates the program output and the quantum circuit structure to preserve a customer’s privacy under a specified computation model and resource availability.
SPYCE [Under Review]
