Yiding Jiang

yd <last name> at cmu dot edu

I am a PhD student at Machine Learning Department of Carnegie Mellon University where I work with Professor Zico Kolter. My research is supported by the Google PhD Fellowship.

Previously, I was an AI Resident at Google Research. I obtained my bachelor of science in Electrical Engineering and Computer Science at UC Berkeley, where I worked on robotics and generative models advised by Professor Ken Goldberg. I have also spent time as a research intern at Meta AI Research and Cerebras Systems.

Google Scholar  /  GitHub  /  Twitter  /  LinkedIn

profile photo
Research

I am interested in understanding how high-capacity machine learning systems built on deep neural networks learn and generalize, and using the insights to improve them further. This entails research on a spectrum of topics including representation learning, reinforcement learning, non-convex optimization, and generalization -- both concrete generalization bounds, and less well-understood empirical phenomena such as out-of-distribution and zero-shot generalization.

Publication

* indicates equal contribution

PontTuset
Understanding prompt engineering may not require rethinking generalization
Victor Akinwande, Yiding Jiang, Dylan Sam, J. Zico Kolter
ICLR, 2024
Sampling and Optimization in Discrete Space workshop, ICML 2023 (Outstanding Paper)
PontTuset
PontTuset
PontTuset
PontTuset
PontTuset Learning Options via Compression
Yiding Jiang*, Evan Z. Liu*, Benjamin Eysenbach, J. Zico Kolter, Chelsea Finn
NeurIPS, 2022
[code]
PontTuset Agreement-on-the-line: Predicting the Performance of Neural Networks under Distribution Shift
Christina Baek, Yiding Jiang, Aditi Raghunathan, J. Zico Kolter
NeurIPS, 2022 (oral)

PontTuset Assessing Generalization of SGD via Disagreement
Yiding Jiang*, Vaishnavh Nagarajan*, Christina Baek, J. Zico Kolter
ICLR, 2022 (spotlight)
[blog post]

PontTuset Methods and Analysis of The First Competition in Predicting Generalization of Deep Learning
Yiding Jiang, Parth Natekar*, Manik Sharma*, Sumukh K Aithal*, Dhruva Kashyap*, Natarajan Subramanyam*,Carlos Lassance*, Daniel M. Roy, Gintare Karolina Dziugaite, Suriya Gunasekar, Isabelle Guyon, Pierre Foret, Scott Yak, Hossein Mobahi, Behnam Neyshabur*, Samy Bengio
PMLR: NeurIPS 2020 Competition and Demonstration Track, 2020
[competiton page]   [Codalab]   [competition dataset]   [competition code]

PontTuset Fantastic Generalization Measures and Where to Find Them
Yiding Jiang*, Behnam Neyshabur*, Hossein Mobahi, Dilip Krishnan, Samy Bengio
ICLR, 2020
"Science meets the Engineering of Deep Learning" workshop, NeurIPS 2019 (oral)

PontTuset Observational Overfitting in Reinforcement Learning
Xingyou Song, Yiding Jiang, Stephen Tu, Yilun Du, Behnam Neyshabur
ICLR, 2020

PontTuset Language as an Abstraction for Hierarchical Deep Reinforcement Learning
Yiding Jiang, Shixiang Gu, Kevin Murphy, Chelsea Finn
NeurIPS, 2019
[project page]   [environment]

PontTuset
PontTuset Predicting the Generalization Gap in Deep Networks with Margin Distributions
Yiding Jiang, Dilip Krishnan, Hossein Mobahi, Samy Bengio
ICLR, 2019
blog post

Technical Report
PontTuset Ask & Explore: Grounded Question Answering for Curiosity-Driven Exploration
Jivat Neet Kaur, Yiding Jiang, Paul Pu Liang
Arxiv, 2021
Workshop on Embodied Multimodal Learning, ICLR 2021

Teaching
  • Teaching Assistant, 10-708 Probablistic Graphical Models. Carnegie Mellon University. Fall 2022.
  • Teaching Assistant, 10-725 Convex Optimization. Carnegie Mellon University. Fall 2021.
  • Reader, CS170 Efficient Algorithms and Intractable Problems. University of California, Berkeley. Fall 2017.
Project
PontTuset CLEVR-Robot Environment
GitHub repository

The CLEVR-Robot environment is a reinforcement learning environment that aims to provide a research platform for developing RL agents at the intersection of vision, language, and continuous/discrete control.

PontTuset Deep Model Generalization Dataset (DEMOGEN)
GitHub repository

The DEMOGEN dataset is a the collection of 756 trained neural network models and the code to use them. This is the same dataset used by our work "Predicting the Generalization Gap in Deep Networks with Margin Distributions", and is to our knoweledge the first dataset of models for studying generalization.

PontTuset City2City
Project Page

City2City is a project that restyles Google streetview images of one city with another city.


Template of this website is made by Jon Barron.