Guan-Horng Liu

Machine Learning PhD @ Georgia Tech 🚀

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, which include, e.g., Neural ODEs, Neural SDEs (diffusion models), interpreting DNNs as discrete-time systems, or mean-field population interaction. The high dimensionality and over-parameterization of these dynamics pose interesting challenges to traditional computational methods. In return, studying these parametrized models from a pure dynamical standpoint enriches theoretical and algorithmic improvements grounded on the first principles of optimal control, stochastic calculus, game theory, optimal transport, and stochastic physics. This new line of research has received spotlight/oral presentations in ICLR’21, ICML’21, and NeurIPS’21.

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

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

Updates

Sep, 2022 :fire: Our Deep Generalized Schrödinger Bridge is accepted to NeurIPS 2022.
Aug, 2022 :mega: I will give an invited talk at Score Models Workshop in NeurIPS 2022!
Jun, 2022 :mega: I have joined Nvidia Research for an internship this summer.
Feb, 2022 :fire: Our Likelihood training of Schrödinger Bridge is accepted to ICLR’22. Code is released here!
Jan, 2022 :books: I am very fortunate to receive GaTech AE Graduate Fellowship. Thanks GaTech AE!
Dec, 2021 :mega: I give a talk on Higher-order Optimization of Neural ODEs at DataSig.
Oct, 2021 :mega: I give a talk on Optimal Control Theoretic Neural Optimizer at GaTech ML PhD Seminar.
Sep, 2021 :fire: Our Second-order Neural ODE Optimizer is accepted to NeurIPS’21 as a Spotlight (3.0%).
May, 2021 :fire: Our Dynamic Game Theoretic Neural Optimizer is accepted to ICML’21 as an Oral (3.0%).
Jan, 2021 :fire: Our Differential Dynamic Programming Neural Optimizer is accepted to ICLR’21 as a Spotlight (3.8%).

Publications

2022

  1. NeurIPS 2022
    Deep Generalized Schrödinger Bridge
    Guan-Horng Liu, Tianrong Chen*, Oswin So*, and Evangelos A Theodorou
    Advances in Neural Information Processing Systems, 2022
  2. ICLR 2022
    Likelihood Training of Schrödinger Bridge using Forward-Backward SDEs Theory
    Tianrong Chen*, Guan-Horng Liu*, and Evangelos A Theodorou
    International Conference on Learning Representations (* Equal contribution), 2022

2021

  1. NeurIPS 2021
    Second-order Neural ODE Optimizer
    Guan-Horng Liu, Tianrong Chen, and Evangelos A Theodorou
    Advances in Neural Information Processing Systems, 2021
    Spotlight presentation (acceptance rate 3.0%)
  2. ICML 2021
    Dynamic Game Theoretic Neural Optimizer
    Guan-Horng Liu, Tianrong Chen, and Evangelos A Theodorou
    International Conference on Machine Learning, 2021
    Oral presentation (acceptance rate 3.0%)
  3. ICLR 2021
    DDPNOpt: Differential Dynamic Programming Neural Optimizer
    Guan-Horng Liu, Tianrong Chen, and Evangelos A Theodorou
    International Conference on Learning Representations, 2021
    Spotlight presentation (acceptance rate 3.8%)
  4. arxiv
    Spatio-Temporal Differential Dynamic Programming for Control of Fields
    Ethan N Evans, Oswin So, Andrew P Kendall, Guan-Horng Liu, and Evangelos A Theodorou
    arXiv preprint (in submission), 2021
  5. RSS 2021
    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, 2021

2020

  1. arxiv
    A Differential Game Theoretic Neural Optimizer for Training Residual Networks
    Guan-Horng Liu, Tianrong Chen, and Evangelos A Theodorou
    arXiv preprint, 2020

2019

  1. arxiv
    Deep learning theory review: An optimal control and dynamical systems perspective
    Guan-Horng Liu, and Evangelos A Theodorou
    arXiv preprint (in submission), 2019

2017

  1. CoRL 2017
    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, 2017
  2. CMU Thesis
    High Dimensional Planning and Learning for Off-Road Driving
    Guan-Horng Liu
    CMU Robotics Institute Master Thesis, 2017
  3. RLDM 2017
    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, 2017

2014

  1. JBE 2014
    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, 2014
  2. RSJ 2014
    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

2013

  1. SII 2013
    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, 2013
    Best Paper Award