Guan-Horng Liu

Machine Learning PhD @ Georgia Tech ūüöÄ

Hi, I am Guan-Horng Liu (I go by "Guan" ), a fourth-year Machine Learning PhD in Georgia Tech advised by Evangelos A. Theodorou. I also work closely with Molei Tao from Georgia Tech and Valentin De Bortoli from CNRS.

My research concerns scalable computational methods for neural dynamics, including Neural ODEs, Neural SDEs (e.g., diffusion models), DNNs (as discrete-time systems), their mean-field extensions, and etc. These neural dynamics pose interesting numerical challenges to traditional computational methods. In return, studies from dynamic and control standpoints have inspired new machine learning pipelines in principled algorithmic design and development of theoretical insights grounded on optimal control, stochastic processes, game theory, dynamic optimal transport, and statistical physics. This new line of research has received spotlight/oral presentations in ICLR’21, ICML’21, NeurIPS’21, and NeurIPS’22.

I am fortunate to work in Nvidia Research during 2022 Summer, developing nonlinear diffusion models with Weili Nie, Arash Vahdat, De-An Huang, and Anima Anandkumar. See my full CV here (updated in April 2023).

Contact: ghliu [at] gatech [dot] edu
Follow: Google Scholar | LinkedIn | ghliu | @guanhorng_liu


Updates
[07/2023] Two papers accepted to ICML Workshop on New Frontiers in Learning, Control, and Dynamical Systems.
[04/2023] One paper, Image-to-Image Schrödinger Bridge, accepts to ICML.
[03/2023] I co-organize the ICML workshop on New Frontiers in Learning, Control, and Dynamical Systems.
[12/2022] I give an invited talks, Generalized Schrödinger Bridge, at NeurIPS Score-based Methods Workshop.
[11/2022] I give two invited talks, Generalized Schrödinger Bridge, at IBM Research and Alan Turing Institute.
[10/2022] I give a guest lecture at AE4803 Robotic Systems and Autonomy.
[09/2022] One paper, Deep Generalized Schrödinger Bridge, accepts to NeurIPS (Oral 1.9%).
[06/2022] I have joined Nvidia Research for an internship this summer.
[02/2022] One paper, Likelihood training of Schrödinger Bridge, accepts to ICLR.
[01/2022] I receive GaTech AE Graduate Fellowship. Thanks GaTech AE!
[12/2021] I give a talk, Higher-order Optimization of Neural ODE, at DataSig, Alan Turing Institute.
[10/2021] I give a talk, Optimal Control Neural Optimizer, at GaTech ML PhD Seminar.
[09/2021] One paper, Second-order Neural ODE Opt, accepts to NeurIPS (Spotlight 3.0%).
[05/2021] One paper, Dynamic Game NOpt, accepts to ICML (Oral 3.0%).
[01/2021] One paper, DDP Neural Optimizer, accepts to ICLR (Spotlight 3.8%).

Publications
(* Equal contribution, † Equal advising)
  1. Game Theoretic Neural ODE Optimizer
    Panagiotis Theodoropoulos, Guan-Horng Liu*, Tiarnong Chen*, and Evangelos A Theodorou
    ICML Workshop on New Frontiers in Learning, Control, and Dynamical Systems, 2023
  2. Improved Sampling via Learned Diffusions
    Lorenz Richter*, Julius Berner*, and Guan-Horng Liu
    ICML Workshop on New Frontiers in Learning, Control, and Dynamical Systems, 2023
  3. I²SB: Image-to-Image Schrödinger Bridge
    Guan-Horng Liu, Arash Vahdat, De-An Huang, Evangelos A Theodorou, Weili Nie†, and Anima Anandkumar†
    International Conference on Machine Learning (ICML), 2023
  4. Deep Momentum Multi-Marginal Schrödinger Bridge
    Tianrong Chen, Guan-Horng Liu, Molei Tao, and Evangelos A Theodorou
    preprint (in submission), 2023
  1. Deep Generalized Schrödinger Bridge
    Guan-Horng Liu, Tianrong Chen*, Oswin So*, and Evangelos A Theodorou
    Advances in Neural Information Processing Systems (NeurIPS), 2022
    Oral presentation (acceptance rate 1.9%)
  2. Likelihood Training of Schrödinger Bridge using Forward-Backward SDEs Theory
    Tianrong Chen*, Guan-Horng Liu*, and Evangelos A Theodorou
    International Conference on Learning Representations (ICLR), 2022
  1. Second-order Neural ODE Optimizer
    Guan-Horng Liu, Tianrong Chen, and Evangelos A Theodorou
    Advances in Neural Information Processing Systems (NeurIPS), 2021
    Spotlight presentation (acceptance rate 3.0%)
  2. Dynamic Game Theoretic Neural Optimizer
    Guan-Horng Liu, Tianrong Chen, and Evangelos A Theodorou
    International Conference on Machine Learning (ICML), 2021
    Oral presentation (acceptance rate 3.0%)
  3. DDPNOpt: Differential Dynamic Programming Neural Optimizer
    Guan-Horng Liu, Tianrong Chen, and Evangelos A Theodorou
    International Conference on Learning Representations (ICLR), 2021
    Spotlight presentation (acceptance rate 3.8%)
  4. Spatio-Temporal Differential Dynamic Programming for Control of Fields
    Ethan N Evans, Oswin So, Andrew P Kendall, Guan-Horng Liu, and Evangelos A Theodorou
    preprint (in submission), 2021
  5. Variational Inference MPC using Tsallis Divergence
    Ziyi Wang, Oswin So, Jason Gibson, Bogdan Vlahov, Manan S Gandhi, Guan-Horng Liu, and Evangelos A Theodorou
    Robotics: Science and Systems (RSS), 2021
    1. Deep learning theory review: An optimal control and dynamical systems perspective
      Guan-Horng Liu, and Evangelos A Theodorou
      preprint (in submission), 2019
    1. Learning end-to-end multimodal sensor policies for autonomous navigation
      Guan-Horng Liu, Avinash Siravuru, Sai Prabhakar, Manuela Veloso, and George Kantor
      Conference on Robot Learning (CoRL), 2017
    2. High Dimensional Planning and Learning for Off-Road Driving
      Guan-Horng Liu
      CMU Robotics Institute Master Thesis, 2017
    3. Multi-modal Deep Reinforcement Learning with a Novel Sensor-based Dropout
      Guan-Horng Liu, Avinash Siravuru, Sai Prabhakar, George Kantor, and Manuela Veloso
      Multi-disciplinary Conference on Reinforcement Learning and Decision Making (MRLD), 2017
    1. A bio-inspired hopping kangaroo robot with an active tail
      Guan-Horng Liu, Hou-Yi Lin, Huai-Yu Lin, Shao-Tuan Chen, and Pei-Chun Lin
      Journal of Bionic Engineering (JBE), 2014
    2. Autonomous Control of the WAM-V Catamaran Type Unmanned Surface Vehicle: Propulsion System Design
      Guan-Horng Liu, Andre Yuji YASUTOMI, Alexis HOLGADO, and Edwardo F FUKUSHIMA
      Annual Conference of the Robotics Society of Japan, 2014
    1. Design of a kangaroo robot with dynamic jogging locomotion
      Guan-Horng Liu, Hou-Yi Lin, Huai-Yu Lin, Shao-Tuan Chen, and Pei-Chun Lin
      Proceedings of the 2013 IEEE/SICE International Symposium on System Integration (SII), 2013
      Best Paper Award