Neural network controllers trained via behavior cloning are increasingly deployed in cyber-physical systems (CPS), yet practitioners lack tools to trace controller failures back to training data. Existing data attribution methods assume i.i.d.\ data and standard loss targets, ignoring CPS-specific properties: closed-loop dynamics, safety constraints, and temporal trajectory structure. We propose IF-CPS, a modular influence function framework with three CPS-adapted variants: safety influence (attributing constraint violations), trajectory influence (temporal discounting over trajectories), and propagated influence (tracing effects through plant dynamics). We evaluate IF-CPS on six benchmarks across diagnosis, curation, and safety attribution tasks. IF-CPS improves over standard influence functions in the majority of settings, achieving AUROC 1.00 in Pendulum (5-10\% poisoning), 0.92 vs.\ 0.50 in HVAC (10\%), and the strongest constraint-boundary correlation (Spearman ρ=0.55 in Pendulum).
@article{arxiv.2603.21543,
title = {IF-CPS: Influence Functions for Cyber-Physical Systems -- A Unified Framework for Diagnosis, Curation, and Safety Attribution},
author = {Jiachen Li and Shihao Li and Soovadeep Bakshi and Jiamin Xu and Dongmei Chen},
journal= {arXiv preprint arXiv:2603.21543},
year = {2026}
}