A few ongoing projects supported by external funds from my group include

• Scalable analog computing for efficient AI/ML inference and training by leveraging approximation error-bounded analog in-memory computing and algorithm co-design;

• Learnable in-sensor compression and multi-modal sensor fusion for end-to-end efficient and high-performance visual processing (e.g., vision-language-action acceleration) and neural signal processing (e.g., cardiac anomaly detection);

• Quantum physics-inspired and agentic AI-driven electronic design automation (EDA) for novel and efficient development of advanced physical AI hardware. 

We are also interested in privacy-preserving machine learning in broad applications, such as data-driven EDA and in-sensor processing.

1) Machine learning for science and engineering, e.g., electronic design automation. We are now exploring machine learning methods to automate the design of circuits and computing systems. For example, we leverage reinforcement learning to automatically size analog/radio-frequency (RF) circuits and use graph learning to generate analog circuit topologies. Please check the prior publications [DAC '22][DAC '23][ICLR '23] for details.

2) Efficient hardware accelerator for machine learning applications. We explore the co-optimization of devices, efficient VLSI, emerging computing architectures (e.g., in- and near-sensor computing and in- and near-memory computing), and algorithms to unleash the performance of machine learning on hardware. Please check my prior publications [DATE '19][ICCAD '19][TCAD '20][TCAD '21][TC '22][ISLPED '22][ISCA '23][ISCAS '23] for details.

3) Robust and scalable quantum computing. We also do fundamental research to enhance the robustness and scalability of superconducting quantum computing systems through the design of circuits and algorithms (e.g., Ising Machine and Topological Electronics). Please check the prior publications [Nature Nanotechnology '22][Scientific Reports '22] for details.