About me

I'm a PhD candidate in Computer Science at Stony Brook University, advised by Dr. Owen Rambow.

My research centers on multimodal grounding and human-AI communication — investigating how Vision-Language Models (LVLMs) ground language to vision and coordinate with human partners, with a focus on where model behavior diverges from human pragmatic inference in interactive settings. More broadly, I'm interested in machine theory of mind: assessing the ability of LLMs to model belief states and track recursive mental states. I'm also interested in AI interpretability, developing hybrid architectures that combine neural network performance with feature-level explainability.

Updates

Publications

  1. Implicit vs. Explicit Prompting Strategies for LVLMs in Referential Communication
    P. Zeng, A. Paige, W. Li, S. Brennan, O. Rambow, C. Jones
    EMNLP 2026 (Main)
  2. LVLMs and Humans Ground Differently in Referential Communication
    P. Zeng, W. Li, A. Paige, Z. Wang, P. Kaliosis, D. Samaras, G. Zelinsky, S. Brennan, O. Rambow
    ACL 2026 (Main)
  3. Residualized Similarity for Interpretability in Authorship Verification
    P. Zeng, P. Alipoormolabashi, J. Mun, G. Dey, N. Soni, N. Balasubramanian, O. Rambow, H. Schwartz
    EMNLP 2025 (Findings)
  4. Synthetic Audio Helps for Cognitive State Tasks
    A. Soubki*, J. Murzaku*, P. Zeng, O. Rambow
    NAACL 2025 (Findings)
  5. Gram2Vec: An Interpretable Document Vectorizer
    P. Zeng, H. Stortz, E. Sclafani, A. Shabaeva, M. E. Garza, D. Greeson, O. Rambow
    Arxiv 2025
  6. Views Are My Own, but Also Yours: Benchmarking Theory of Mind Using Common Ground
    A. Soubki, J. Murzaku, A. Y. Jordehi, P. Zeng, M. Markowska, S. A. Mirroshandel, O. Rambow
    ACL 2024 (Findings)
  7. Re-Examining Factbank: Predicting the Author's Presentation of Factuality
    J. Murzaku, P. Zeng, M. Markowska, O. Rambow
    COLING 2022

Contact