Tianyi Liu

Ph.D Student in Operations Research

Georgia Institute of Technology


Email: tianyiliu(at)gatech(dot)edu

Short Bio

I am a PhD student in the School of Industrial and Systems Engineering (ISyE) at Georgia Tech. Before I joined Georgia Tech, I received my B.S. Degree in Mathematics from Nanjing University, China, in 2016.

My research focuses on non-convex optimization in deep learning and simulation. I am currently working with Prof. Enlu Zhou and Prof. Tuo Zhao in the FLASH (Foundations of LeArning Systems for alcHemy) research group.

Publications

(Note: An asterisk * indicates equal contribution.)

Non-convex Optimization and Deep Learning

  • Tianyi Liu*, Minshuo Chen*, Mo Zhou, Simon Du, Enlu Zhou and Tuo Zhao. " Towards Understanding the Importance of Shortcut Connections in Residual Networks." Annual Conference on Neural Information Processing Systems (NeurIPS' 2019). [arXiv]

  • Mo Zhou*, Tianyi Liu*, Yan Li, Dachao Lin, Enlu Zhou and Tuo Zhao. "Towards Understanding the Importance of Noise in Training Neural Networks ." International Conference on Machine Learning (ICML' 2019) (Long Oral). [arXiv]

  • Tianyi Liu, Shiyang Li, Jianping Shi, Enlu Zhou and Tuo Zhao . "Towards Understanding Acceleration Tradeoff between Momentum and Asynchrony in Distributed Nonconvex Stochastic Optimization ." Annual Conference on Neural Information Processing Systems (NeurIPS' 2018). [arXiv]

  • Minshuo Chen, Yizhou Wang, Tianyi Liu, Zhuorang Yang, Xingguo Li, Zhaoran Wang, and Tuo Zhao. "On Computation and Generalization of Generative Adversarial Imitation Learning." International Conference on Learning Representations (ICLR' 2020).[pdf]

  • Tianyi Liu, Zhehui Chen, Enlu Zhou and Tuo Zhao. "Towards Deeper Understanding of Nonconvex Stochastic Optimization with Momentum using Diffusion Approximations." Submit to Stochastic Systems. [arXiv]

  • Guojing Cong, Brian Kingsbury, Chih-Chieh Yang, Tianyi Liu. "Fast Training of Deep Neural Networks for Speech Recognition." IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2020. [IEEE]

Simulation

  • Helin Zhu, Tianyi Liu, and Enlu Zhou. "Risk Quantification in Stochastic Simulation under Input Uncertainty ." ACM Transactions on Modeling and Computer Simulation, 2019. [arXiv]

  • Enlu Zhou and Tianyi Liu. "Online Quantification of Input Uncertainty for Parametric Models" (short version). Winter Simulation Conference, 2018. [Paper]

  • Tianyi Liu, Enlu Zhou. "Simulation Optimization by Reusing Past Replications: Don’t Be Afraid of Dependence." Winter Simulation Conference (WSC), 2020

  • Tianyi Liu, Enlu Zhou. "Online Quantification of Input Model Uncertainty by Two-Layer Importance Sampling." Submitted to Operations Research, under major revision. [arXiv]