Tenured associate professor of information at the University of Michigan.

Co-founder and VP of technology of the 501c3 non-profit organization I-Connect Foundation (https://www.iconnectfoundation.org/), use state-of-the-art AI technologies to support socially isolated seniors and detect early dementia.

Research on big data, machine learning and biomedical informatics.

• Leading the machine learning and big data analytics research.• Designing and developing low-power deep learning models and learning algorithms.

• Led the design and development of a large-scale recommender system based on Apache Spark framework. The recommender system learns users' preferences from terabytes of daily watch histories, extracts item profiles using natural language processing technologies, combines them with advanced machine learning models, and delivers personalized recommendations.• Designed novel hybrid algorithms for recommender system to address the cold-start program recommendation.

Research directions: machine learning, data mining, biomedical informatics.• Identified stable biomarkers for prognosis of probable Alzheimer’s disease from MRI, PET, CSF and plasma panel.• Built general disease progression models using multi-task learning techniques. The disease progression models were successfully applied to the prediction of Alzheimer’s disease and cancer progression.• Developed efficient optimization algorithms for solving multi-task learning formulations.• Led the development of the MALSAR package for multi-task learning, including solvers for most state-of-the-art multi-task learning formulations. The package is available at http://www.MALSAR.org

• Developed and implemented a language translation system from natural language to formal languages using sparse learning methods.

• Fall 2010 CSE205: Java Programming Language• Spring 2010 CSE230: Assembly Language and Computer Architecture• Fall 2009 CSE205: Java Programming Language

Designed machine learning models for effective and efficient electronic medical record (EMR) imputation for large scale healthcare problems, leveraging the temporal nature of the patients and the shared information among the patients. The models significantly improved the predictive performance of healthcare systems.

• Designed and implemented a sparse predictive modeling pipeline that takes advantage of the sparsity in the data matrices and has greatly improved the scalability of predictive modeling. • Designed experiments to build effective hospitalization admission models and identify important clinical features for congested heart failure patients from their health records.• Devised a model and algorithm for simultaneously identifying dynamic biomarker and building predictive models (1 patent in progress). • Devised a model and algorithm for learning multiple tasks with graphical outcome constraints (1 patent in progress).

Research direction: natural language processing, machine translation. • Developed and implemented a rule-based machine translation system from English to Chinese.

Research direction: information retrieval, search behavior, data mining.• Analyzed large-scale log data of Microsoft Live search engine and modeled user behavior patterns to improve search experience.

Research direction: knowledge engineering, text mining, natural language processing.Developed and published a Chinese concept mining system. Developed and published a hyponymy detection system.