Xin Du
Associate Professor
Shanghai Research Institute for Intelligent Autonomous Systems
Tongji University
I have published more than ten first-author papers at ICML, NeurIPS, AAAI, ACL, and in peer-reviewed journals. My ICML 2024 paper was selected for an oral presentation.
Appointments
- 2026–present: Associate Professor (tenured), Tongji University
- 2023–2026: Assistant Professor, Waseda University
- 2021–2023: JSPS Research Fellow
Education
- 2020–2023: PhD in Engineering (Advanced Interdisciplinary Studies), The University of Tokyo
- 2018–2020: MS in Information Science (Mathematical Informatics), The University of Tokyo
- 2012–2017: Bachelor of Engineering, Tongji University
Research
I study the dynamics and complexity of foundation models and AI agents, with a focus on long-term stability, semantic compression, and efficient sequence representations. I also develop semantic models of financial markets for event analysis and portfolio risk assessment.
Agent Dynamics
Instability, order parameters, and early-warning signals in long-horizon generation and agent loops, with online regulation of internal states.
ICML 2026 02 · COMPLEXITYFoundation Models and Complexity
Correlation dimension on statistical manifolds as a probe of pretraining phases, long-range structure, hallucination, and degeneration.
NeurIPS 2025 · PRR 2024 03 · COMPRESSIONSemantic Compression and Clustering
Information bottlenecks and conditional generative distributions for discrete indexing, document clustering, and generative retrieval.
ICML 2024 Oral · AAAI 2025 04 · REPRESENTATIONNonlinear Semantics and Linear Language Models
Function-valued semantic fields for polysemy, and selection, writing, and finite-state memory in Mamba, DeltaNet, and related models.
NeurIPS 2022 05 · FINANCESemantic Models of Financial Markets
Joint representations of news, events, and prices for stock embeddings, portfolio optimization, and tail-risk analysis.
ACL 2020 · KBS 2022