Zhang Junyu

Research Interest:

• Saddle point problems algorithm design & analysis, complexity lower bounds

• Stochastic optimization: variance reduction methods, sample complexity analysis

• Optimization theories of Reinforcement Learning

• Composite optimization: prox-linear methods, stochastic composite optimization

• Riemannian optimization

Selected Journal Paper:

• Bedi, AS., Parayil, A., Zhang, J., Wang, M., Koppel, A. 2024. On the sample complexity and metastability of heavy-tailed policy search in continuous control. Journal of Machine Learning Research, 25 (39), 1-58. [paper]

• Zhang, J., Wang, M., Hong, M., & Zhang, S. 2023. Primal-dual first-order methods for affinely constrained multi-block saddle point problems. SIAM Journal on Optimization, Accepted, to appear soon. [paper]

• Hong, M., Zeng, S., Zhang, J., & Sun, H., 2022. On the Divergence of Decentralized Non-Convex Optimization. SIAM Journal on Optimization, 2022, 32(4): 2879-2908. [paper]

• Zhang, J., Hong, M. and Zhang, S., 2022. On lower iteration complexity bounds for the saddle point problems. Mathematical Programming. 2022, 194(1-2): 901-935. [paper]

• Zhang, J., & Xiao, L., 2021. Stochastic variance-reduced prox-linear algorithms for nonconvex composite optimization. Mathematical Programming. 2021, 1-43. [paper]

• Davis, D., Drusvyatskiy, D., Xiao, L. and Zhang, J., 2021. From Low Probability to High Confidence in Stochastic Convex Optimization. Journal of Machine Learning Research, 22, pp.49-1. [paper]

• Zhang, J. and Xiao, L., 2021. MultiLevel Composite Stochastic Optimization via Nested Variance Reduction. SIAM Journal on Optimization, 31(2), pp.1131-1157. [paper]

• Zhang, J., Ma, S. and Zhang, S., 2020. Primal-dual optimization algorithms over Riemannian manifolds: an iteration complexity analysis. Mathematical Programming, 184(1), pp.445-490. [paper]

• Zhang, J., Liu, H., Wen, Z. and Zhang, S., 2018. A sparse completely positive relaxation of the modularity maximization for community detection. SIAM Journal on Scientific Computing, 40(5), pp.A3091-A3120. [paper]

Selected Conference Proceedings:

• Chen, F., Zhang, J., Wen Z., 2022, A Near-Optimal Primal-Dual Method for Off-Policy Learning in CMDP. Advances in Neural Information Processing Systems (NeurIPS). [paper]

• Zhang, J., Ni, C., Yu, Z., Szepesvari, C., & Wang, M., 2021. On the convergence and sample efficiency of variance-reduced policy gradient method. Advances in Neural Information Processing Systems (NeurIPS).  [paper] (selected as spotlight paper)

• Zhang, J., Hong, M., Wang, M. and Zhang, S., 2021, March. Generalization bounds for stochastic saddle point problems. International Conference on Artificial Intelligence and Statistics (AISTATS), pp. 568-576. PMLR. [paper]

• Zhang, J., Koppel, A., Bedi, A.S., Szepesvari, C. and Wang, M., 2020. Variational policy gradient method for reinforcement learning with general utilities. Advances in Neural Information Processing Systems (NeurIPS), 33, pp.4572-4583. [paper]  (selected as spotlight paper)

• Zhang, J. and Xiao, L., 2019, May. A composite randomized incremental gradient method. International Conference on Machine Learning (ICML), pp. 7454-7462, PMLR. [paper]

• Zhang, J. and Xiao, L., 2019. A stochastic composite gradient method with incremental variance reduction. Advances in Neural Information Processing Systems (NeurIPS), 32, pp.9078-9088. [paper]