In this proposal, we aim to develop innovative and fast online learning algorithms for overcoming limitations suffered by existing and traditional algorithms. In particular, the specific aims of this project include: (i) developing effective online learning algorithms attempting to optimize various asymmetric measures including the F-score, area under ROC curve and area under precision and recall curve; (ii) developing effective online learning algorithms under various constraints on computation, memory, resources and performance; and (iii) evaluating the proposed algorithms in real applications such as biomedical semantic indexing and social media mining. The educational goals of this project are (i) to train future professionals in big data analytics with expertise in dealing with large-scale streaming, heterogeneous and unstructured data; and (ii) to introduce the basic concepts of big data and online machine learning to undergraduate and high school students.

Students

1. Xiaoxuan Zhang (joined the project since 2015 September)
2. Zhe Li (joined the project since 2015 September)
3. Mingrui Liu (joined the project since 2016 August)

Publications

1. Lijun Zhang, Tianbao Yang, Rong Jin. Empirical Risk Minimization for Stochastic Convex Optimization: O(1/n)- and O(1/n^2)-type of Risk Bounds. International Conference on Learning Theory, 2017 (PDF)
2. Tianbao Yang, Qihang Lin, Lijun Zhang. A Richer Theory of Convex Constrained Optimization with Reduced Projections and Improved Rates. International Conference on Machine Learning, 2017 (PDF)
3. Yichi Xiao, Zhe Li, Tianbao Yang, Lijun Zhang. SVD-free Convex-Concave Approaches for Nuclear Norm Regularization. International Joint Conference on Artificial Intelligence, 2017 (PDF)
4. Yi Xu, Qihang Lin, Tianbao Yang. Stochastic Convex Optimization: Faster Local Growth Implies Faster Global Convergence. International Conference on Machine Learning, 2017 (PDF)
5. Yi Xu, Haiqing Yang, Lijun Zhang, Tianbao Yang. Efficient Non-oblivious Randomized Reduction for Risk Minimization with Improved Excess Risk Guarantee. AAAI Conference on Artificial Intelligence, 2017 (PDF)
6. Yan Yan, Tianbao Yang, Yi Yang, Jianhui Chen. A Framework of Online Learning with Imbalanced Streaming Data. AAAI Conference on Artificial Intelligence, 2017 (PDF)
7. Zhe Li, Tianbao Yang, Lijun Zhang, Rong Jin. A Two-stage Approach for Learning a Sparse Model with Sharp Excess Risk Analysis. AAAI Conference on Artificial Intelligence, 2017 (PDF)
8. Xiaoxuan Zhang, Tianbao Yang, Padmini Srinivasan. Online Asymmetric Active Learning with Imbalanced Data. ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD), 2016 (PDF)
9. Tianbao Yang, Lijun Zhang, Rong Jin, Jinfeng Yi. Tracking Slowly Moving Clairvoyant: Optimal Dynamic Regret of Online Learning with True and Noisy Gradient. International Conference on Machine Learning, 2016 (PDF)
10. Lijun Zhang, Tianbao Yang, Rong Jin, Zhi-Hua Zhou. Sparse Learning for Large-scale and High-dimensional Data: A Randomized Convex-concave Optimization Approach. The Conference on Algorithmic Learning Theory (ALT), 2016 (PDF)
11. Jianhui Chen, Tianbao Yang, Qihang Lin, Lijun Zhang, Yi Chang. Optimal Stochastic Strongly Convex Optimization with a Logarithmic Number of Projections. The Conference on Uncertainty in Artificial Intelligence (UAI), 2016 (PDF)
12. Lijun Zhang, Tianbao Yang, Rong Jin, Zhi-Hua Zhou. Online Stochastic Linear Optimization under One-bit Feedback. International Conference on Machine Learning (ICML), 2016 (PDF)
13. Chuang Guan, Tianbao Yang, Boqing Gong. Learning Attributes Equals Multi-Source Domain Generalization. The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016 (PDF)
14. Lijun Zhang, Tianbao Yang, Rong Jin, Zhi-Hua Zhou, Stochastic Optimization for Kernel PCA, AAAI Conference on Artificial Intelligence, 2016 (PDF)
15. Zhe Li, Tianbao Yang, Lijun Zhang, Rong Jin, Fast and Accurate Refined Nystrom based Kernel SVM, AAAI Conference on Artificial Intelligence, 2016 (PDF)
16. Tianbao Yang, Rong Jin, Shenghuo Zhu, Qihang Lin. On Data Preconditioning for Regularized Loss Minimization. Machine Learning Journal, 2015(PDF)
17. Zhe Li, Tianbao Yang, Lijun Zhang, Rong Jin, A two-stage approach for learning a sparse model with sharp excess risk analysis, In NIPS workshop on Learning faster from easy data, 2015 (PDF)
18. Adams Wei Yu, Qihang Lin, Tianbao Yang, Doubly Stochastic Primal-Dual Coordinate Method for Regularized Empirical Risk Minimization with Factorized Data, In NIPS workshop on Optimization for Machine Learning, 2015. (PDF)