Niki van Stein

2025

Scalable, Explainable and Provably Robust Anomaly Detection with One-Step Flow Matching

Zhong Li, Qi Huang, Yuxuan Zhu, Lincen Yang, Mohammad Mohammadi Amiri, Niki van Stein +1

NeurIPS 2025poster

We introduce Time-Conditioned Contraction Matching (TCCM), a novel method for semi-supervised anomaly detection in tabular data. TCCM is inspired by flow matching, a recent generative modeling framework that learns velocity fields between probability distributions and has shown strong performance co…

Cited by 0 Source ↗code ↗Scholar ↗

How Does Chain of Thought Think? Mechanistic Interpretability of Chain-of-Thought Reasoning with Sparse Autoencoding

Scalable, Explainable and Provably Robust Anomaly Detection with One-Step Flow Matching