research

publications by categories in reversed chronological order. generated by jekyll-scholar.

My current research interests span the following:

  • understanding training dynamics of deep models: particularly with respect to factors such as implicit bias, random variation, and scale
  • finetuning language models: preference learning and RLHF
  • reinforcement learning

I’ve also had the privilege of previously working on research projects in theoretical computer science and applied machine learning. See a list of publications below, or check out my Google Scholar for a more up-to-date list.

2024

  1. Policy Gradient Methods in the Presence of Symmetries and State Abstractions
    Panangaden, Prakash, Rezaei-Shoshtari, Sahand, Zhao, Rosie, Meger, David, and Precup, Doina
    Journal of Machine Learning Research 2024
  2. Beyond Implicit Bias: The Insignificance of SGD Noise in Online Learning
    Vyas, Nikhil, Morwani, Depen, Zhao, Rosie, Kaplun, Gal, Kakade, Sham M, and Barak, Boaz
    In Forty-first International Conference on Machine Learning 2024
  3. Feature emergence via margin maximization: case studies in algebraic tasks
    Morwani, Depen, Edelman, Benjamin L, Oncescu, Costin-Andrei, Zhao, Rosie, and Kakade, Sham M
    In The Twelfth International Conference on Learning Representations 2024
  4. Deconstructing What Makes a Good Optimizer for Language Models
    Zhao, Rosie, Morwani, Depen, Brandfonbrener, David, Vyas, Nikhil, and Kakade, Sham
    arXiv preprint arXiv:2407.07972 2024
  5. SOAP: Improving and Stabilizing Shampoo using Adam
    Vyas, Nikhil, Morwani, Depen, Zhao, Rosie, Shapira, Itai, Brandfonbrener, David, Janson, Lucas, and Kakade, Sham
    arXiv preprint arXiv:2409.11321 2024

2023

  1. On the peel number and the leaf-height of Galton–Watson trees
    Devroye, Luc, Goh, Marcel K, and Zhao, Rosie Y
    Combinatorics, Probability and Computing 2023
  2. Loss of plasticity in continual deep reinforcement learning
    Abbas, Zaheer, Zhao, Rosie, Modayil, Joseph, White, Adam, and Machado, Marlos C
    In Conference on Lifelong Learning Agents 2023

2022

  1. Leaf multiplicity in a Bienaym\backslash’e-Galton-Watson tree
    Brandenberger, Anna M, Devroye, Luc, Goh, Marcel K, and Zhao, Rosie Y
    Discrete Mathematics & Theoretical Computer Science 2022
  2. Boolean functions with small approximate spectral norm
    Cheung, Tsun Ming, Hatami, Hamed, Zhao, Rosie, and Zilberstein, Itai
    In 2022
  3. Lower bound methods for sign-rank and their limitations
    Hatami, Hamed, Hatami, Pooya, Pires, William, Tao, Ran, and Zhao, Rosie
    In Approximation, Randomization, and Combinatorial Optimization. Algorithms and Techniques (APPROX/RANDOM 2022) 2022
  4. Continuous mdp homomorphisms and homomorphic policy gradient
    Rezaei-Shoshtari, Sahand, Zhao, Rosie, Panangaden, Prakash, Meger, David, and Precup, Doina
    Advances in Neural Information Processing Systems 2022
  5. Continuous Homomorphisms and Leveraging Symmetries in Policy Gradient Algorithms for Markov Decision Processes
    Zhao, Rosie Y
    2022

2021

  1. Bridging the gap between supervised classification and unsupervised topic modelling for social-media assisted crisis management
    Brunila, Mikael, Zhao, Rosie, Mircea, Andrei, Lumley, Sam, and Sieber, Renee
    In Proceedings of the Second Workshop on Domain Adaptation for NLP 2021
  2. Arithmetic Subsequences in a Random Ordering of an Additive Set
    Goh, Marcel K, and Zhao, Rosie Y
    INTEGERS 2021

2020

  1. Using deep learning and social network analysis to understand and manage extreme flooding
    Romascanu, Andrei, Ker, Hannah, Sieber, Renee, Greenidge, Sarah, Lumley, Sam, Bush, Drew, Morgan, Stefan, Zhao, Rosie, and Brunila, Mikael
    Journal of Contingencies and Crisis Management 2020