Selected talks

  • Decentralized Methods for Composite Nonconvex Strongly-Concave Minimax Problems. SIAM-NNP Annual Meeting, Oct. 21, 2023.

  • Lower complexity bounds of first-order methods for affinely constrained nonconvex nonsmooth problems. INFORMS Annual Meeting, Oct. 17, 2023

  • Inexact Proximal Augmented Lagrangian Method for weakly-Convex Problems with Convex Constraints. MOPTA, August 18, 2023.

  • Compressed Proximal Stochastic Gradient for Decentralized Composite Nonconvex Optimization. SIAM Conference on Optimization, June 01, 2023.

  • Decentralized gradient descent maximization method for composite nonconvex strongly-concave minimax problems. Lehigh University ISE Department Seminar, March 07, 2023.

  • Low-order Methods for Nonlinear Constrained Nonconvex Optimization. INFORMS Annual Meeting, Indianapolis, IN, Oct. 16, 2022.

  • First-order methods for nonlinear optimization with a few constraints, ICCOPT, Lehigh University, July 26, 2022. [Slides]

  • Distributed stochastic inertial-accelerated methods with delayed derivatives, INFORMS Optimization Society Conference, Greenville, SC, March 13, 2022. [Slides]

  • First-order methods for nonlinear-constrained optimization, Machine Learning Seminar, UMN, Virtual, Dec. 02, 2021.

  • Asynchronous Parallel Adaptive Stochastic Gradient Methods, SIAM Conference on CSE, Virtual, Mar. 04, 2021.

  • Near-optimal first-order methods for nonlinear programs with O(1) functional constraints, PKU Workshop on Optimization Theory and Methods, Virtual, Jan. 31, 2021.

  • Adaptive primal-dual stochastic gradient methods, INFORMS Annual Meeting, Virtual, Nov. 13, 2020. [Slides]

  • Lower complexity bounds of first-order methods for convex-concave bilinear saddle-point problems, ICCOPT, Berlin, August 06, 2019. [Slides]

  • Asynchronous parallel primal-dual block update method. MOPTA, Lehigh University, Aug. 16–18, 2017. [Slides]

  • Accelerated primal-dual methods for linearly constrained convex problems. SIAM Conference on Optimization. Vancouver, Canada, May 22–25, 2017. [Slides]