Wonyeol Lee

Incoming Assistant Professor (November 2024)
Computer Science Department, POSTECH

Postdoctoral Associate
Computer Science Department, CMU

wonyeol.lee.cs@gmail.com  |  CV

I am recruiting motivated and talented students at all levels.
If you are interested, please email me with your CV, transcript, and research interests.

About Me

I am a Postdoctoral Associate at CMU, working with Feras Saad. I received my Ph.D. in Computer Science from Stanford University, working under Alex Aiken. During my Ph.D., I also spent time at KAIST for military service, working with Hongseok Yang. Before that, I obtained my B.S. degree in Computer Science and Mathematics from POSTECH.

My research aims at making continuous computations more reliable and more scalable. Towards this goal, I consider a wide range of continuous computations arising in different areas (e.g., programming languages and machine learning), and study their correctness and efficiency in three main directions:

• Analyze existing, practical continuous computations from theoretical perspectives.

• Design new, practical continuous computations that have theoretical guarantees.

• Understand the fundamental limits of continuous computations.

Research Interests

• Continuous Computations

  • - Continuous (floating point | math library | neural network)

  • - Differentiable (non-differentiability | automatic differentiation | gradient estimation)

  • - Probabilistic (sample generation | variational inference)

• Correctness & Efficiency

  • - Provable Guarantee (verification | program analysis)

Publications

1. Expressive Power of ReLU and Step Networks under Floating-Point Operations
   Yeachan Park, Geonho Hwang, Wonyeol Lee, Sejun Park
   Preprint, 2024

2. What Does Automatic Differentiation Compute for Neural Networks?
   Sejun Park, Sanghyuk Chun, Wonyeol Lee
   ICLR 2024 (Spotlight)

3. Reasoning About Floating Point in Real-World Systems
   Wonyeol Lee
   Ph.D. Dissertation, 2023
   slides

4. On the Correctness of Automatic Differentiation for Neural Networks with Machine-Representable Parameters
   Wonyeol Lee, Sejun Park, Alex Aiken
   ICML 2023
   slides

5. Training with Mixed-Precision Floating-Point Assignments
   Wonyeol Lee, Rahul Sharma, Alex Aiken
   TMLR 2023
   code

6. Smoothness Analysis for Probabilistic Programs with Application to Optimised Variational Inference
   Wonyeol Lee, Xavier Rival, Hongseok Yang
   POPL 2023
   slides (long) | video | code

7. On Correctness of Automatic Differentiation for Non-Differentiable Functions
   Wonyeol Lee, Hangyeol Yu, Xavier Rival, Hongseok Yang
   NeurIPS 2020 (Spotlight)
   slides (long) | video

8. Differentiable Algorithm for Marginalising Changepoints
   Hyoungjin Lim, Gwonsoo Che, Wonyeol Lee, Hongseok Yang
   AAAI 2020

9. Towards Verified Stochastic Variational Inference for Probabilistic Programs
   Wonyeol Lee, Hangyeol Yu, Xavier Rival, Hongseok Yang
   POPL 2020
   slides | video | code

10. Reparameterization Gradient for Non-Differentiable Models
    Wonyeol Lee, Hangyeol Yu, Hongseok Yang
    NeurIPS 2018
    slides | code

11. On Automatically Proving the Correctness of math.h Implementations
    Wonyeol Lee, Rahul Sharma, Alex Aiken
    POPL 2018
    slides (short) | video

12. Verifying Bit-Manipulations of Floating-Point
    Wonyeol Lee, Rahul Sharma, Alex Aiken
    PLDI 2016
    slides | video

13. A Proof System for Separation Logic with Magic Wand
    Wonyeol Lee, Sungwoo Park
    POPL 2014

14. CT-IC: Continuously Activated and Time-Restricted Independent Cascade Model for Viral Marketing
    Wonyeol Lee, Jinha Kim, Hwanjo Yu
    ICDM 2012
    journal | slides

15. Edge Detection Using Morphological Amoebas in Noisy Images
    Wonyeol Lee, Seyun Kim, Youngwoo Kim, Jaeyoung Lim, Dong Hoon Lim
    ICIP 2009
    journal

trips | last updated: 03/2024