Scholar
Ziwei Xu
Google Scholar ID: 4LROGysAAAAJ
National University of Singapore
Citations & Impact
Citations
1,130
H-index
9
i10-index
9
Publications
19
Co-authors
17
list available
Publications
Resume (English only)
Academic Achievements
- Published multiple papers in top-tier venues including NeurIPS, ICLR, CVPR, and IEEE Transactions on Multimedia.
- Notable publications include:
- – 'Reasoning LLMs are Wandering Solution Explorers' (2025, preprint)
- – 'Bullying the Machine: How Personas Increase LLM Vulnerability' (2025, preprint)
- – 'Strong Preferences Affect the Robustness of Value Alignment' (ICLR 2025)
- – 'Hallucination is Inevitable: An Innate Limitation of Large Language Models' (2024, preprint)
- – 'Bilateral Adaptation for Human-Object Interaction Detection with Occlusion-Robustness' (CVPR 2024)
- – 'Don’t Pour Cereal into Coffee: Differentiable Temporal Logic for Temporal Action Segmentation' (NeurIPS 2022)
- – 'Unsupervised Motion Representation Learning with Capsule Autoencoders' (NeurIPS 2021)
- – 'Relation-aware Compositional Zero-shot Learning for Attribute-Object Pair Recognition' (IEEE TMM 2021)
- – 'Embedding Symbolic Knowledge into Deep Networks' (NeurIPS 2019)
Background
- Machine Learning Engineer at TikTok/ByteDance and an independent researcher.
- Research focuses on: (1) knowledge-enhanced machine learning, integrating symbolic and formal methods with ML models; (2) safety and trustworthiness of AI models.
- Aims to better understand real-world complexities and build more robust, interpretable, and trustworthy models.
- Interests include sequential relation modeling, such as visual relation detection, video understanding, and natural language processing.
- Also explores how powerful AI models impact society and human cognition.
