Formulating few-step generation as a controlled base generative proceses shows that self-consistency loss can be understood through the lens of optimal control. This perspective motivates its generalization to the proposed cumulative self-consistency loss that cumulatively penalizes misalignment along the entire trajectory (ICLR2026) 
Hierarchical Bayesian inference framework for PFL by deriving a personalized posterior distribution over the parameters of each client’s local model (ICML2025) 
Novel model selection method to automate choice of deep neural network architectures (NeurIPS2021, NeurIPS2023, ICML2024) 
Seek innovations in extracting, understanding, and fusing tacit human domain knowledge to advance image semantics interpretation in knowledge rich domains (NeurIPS2020, AAAI2021) 
Aggregate gene expressions with context-dependent topological structures of gene regulations to infer new relationships (NeurIPS2019, BMC System Biology) 
Development and validation of a computational system that is able to automatically identify deceptive intent during computer-mediated, face-to-face communication (FG2015, AAAI2018) 
Understanding the capabilities of the human visual system with respect to biomedical imaging and in extracting and utilizing tacit knowledge of domain experts (CVPR2013, IJCAI2017, CVIU) 