Selena (Ruiqi) Ge


Master Student at UCSD
sege@ucsd.edu / Github / LinkedIn / CV

I am a first year Master student of Data Science at the University of California, San Diego (UCSD), advised by Prof. Biwei Huang. Before that, I was an undergraduate student of Mathematics at New York University (NYU) Shanghai, where I was fortunate to also work with Prof. Mathieu Laurière, Prof. Roberto Fernandez, Prof. Loucas Pillaud-Vivien, and Prof. Jiding Zhang.

My research interests include Diffusion Models, Machine Learning(ML), Reinforcement Learning(RL), Computer Vision(CV), Multimodal Models, Stochastic Processes, Causality. I am passionate about bridging the gap between the elegance of mathematics and the practical demands of artificial intelligence in order to solve real-world downsteam tasks such as controllable image generation.

News

  • 2025-03: My first work on Reinforcement Learning: "Ada-Diffuser: A Causal Diffusion Framework for Latent Identification in RL".
  • 2024-09: Started my Master study of Data Science at UCSD, Halıcıoğlu Data Science Institute (HDSI)!
  • 2024-05: Graduated with Major Honors, NYU Founder's Day Award, Latin Honors: Cum Laude from NYU and NYU Shanghai.

    • Publications and Symposiums

    Ada-Diffuser: A Causal Diffusion Framework for Latent Identification in RL
    Fan Feng, Selena Ge, , Minghao Fu, Zijian Li, Yujia Zheng, Zeyu Tang, Yingyao Hu, Biwei Huang, Kun Zhang
    In submission to NeurIPS 2025
    Bayesian Modeling of Information Spillovers in Post-Pandemic Travel Decisions
    Ruiqi Ge, Jiaqi Liu
    Advisor: Prof. Jiding Zhang
    Oral Presentation at QUIS18 2023, and NYU Shanghai DURF Symposium 2022
    Poster | Slides

    Selected Research

    Enhance Cross-Cultural Toxic Speech Detection with Concept Bottleneck LLMs
    Independent Study project supervised by Prof. Lily Weng, which extended Concept Bottleneck LLMs (CB-LLM) to multilingual toxic speech detection, targeting fairness and interpretability across cultural and linguistic contexts.
    Paper | Slides
    Stochastic Differential Equations(SDE) for Large-scale Machine Learning
    NYU Undergraduate Thesis supervised by Prof. Mathieu Laurière and Prof. Roberto Fernandez. Clarified the connection between SDE and two cutting-edged diffusion models, demonstrating how score-based methods can reverse the noise addition process inherent in diffusion models.
    Paper | Slides
    Modeling Stochastic Gradient Descent(SGD) with Continuous-Time Dynamics
    Independent Study project supervised by Prof. Loucas Pillaud-Vivien. Mathematically proved that SGD learning dynamic is equivalent to a multi-dimensional Ornstein-Uhlenbeck process by modeling the discrete iterative process as a continuous-time SDE. Also validated the theoretical model through numerical simulations, implementing both the SGD algorithm and the corresponding SDE model via the Euler-Maruyama method.
    Paper | Slides

    Teaching Experience

    • Linear Algebra (MATH-SHU 140), Spring 2022, at NYU Shanghai

    • Pre-Calculus (MATH-SHU 9), Fall 2021, at NYU Shanghai

    • Calculus I (MATH-UA 140), Fall 2021, at NYU (remote)

    Leadership: As the Lead Learning Assistant (LA, i.e. undergraduate teaching assistant), I was entrusted with promoting the pedagogical development within the mathematics department, with a mission to enhance peer-led academic support. I organized several large-scale Academic Resource Center events, training sessions, and workshops, and spearheaded the first implementation of LA-supported instruction at NYU main campus.

    © 2025 Selena Ge. Template from Nicklas Hansen. No web trackers.