About
I am a graduate student in Electrical and Computer Engineering at Georgia Tech. My work focuses on building reliable and data-efficient AI systems with strong representation learning.
My research focuses on reliable and data-efficient AI systems, with an emphasis on representation learning and controllability. I am broadly interested in methods that enable robust generalization and efficient adaptation in interactive and language-based settings.
Research Focus
-
LLM Persona Modeling
Understanding and controlling persona-level representations in LLMs.
-
World Models
Learning predictive world models for robust embodied intelligence.
-
Semantic Hashing
Learning compact semantic representations for efficient retrieval.
-
Robust AI Systems
Building AI systems that remain reliable under distribution shift.
Selected Publications
-
X Ma, R Li, X Zhang, Z Weng. "UniHash: Unifying Pointwise and Pairwise Hashing Paradigms for Seen and Unseen Category Retrieval."
arXiv preprint, 2026 -
X Ma, X Zhang, Z Weng. "Stable and Explainable Personality Trait Evaluation in Large Language Models with Internal Activations."
arXiv preprint, 2026 -
G Zollicoffer, T Chopra, M Yan, X Ma, K Eaton, M Riedl. "World Model Robustness via Surprise Recognition."
arXiv preprint, 2025 -
C Ye, S Shang, X Ma, X Zhang. "Input-Envelope-Output: Auditable Generative Music Rewards in Sensory-Sensitive Contexts."
CHI 2026 (Accepted)Archive: to be released.
Academic Service
Reviewer: ACL Rolling Review (ARR), January 2026
Mentoring: I am happy to advise students from underrepresented backgrounds on graduate applications, research planning, and academic growth.