arXiv:2603.25670v3 Announce Type: replace Abstract: Safety monitoring is essential for Cyber-Physical Systems (CPSs). However, unsafe events are rare in real-world CPS operations, creating an extreme class imbalance that degrades safety predictors. Standard rebalancing techniques perform poorly on time-series CPS telemetry, either generating unrealistic synthetic samples or overfitting on the minority class. Meanwhile, behavioral uncertainty in CPS operations, defined as the degree of doubt or uncertainty in CPS decisions , is often correlated with safety outcomes but unexplored in safety moni
The increasing complexity and integration of AI into Cyber-Physical Systems necessitate robust safety monitoring, pushing research towards more effective anomaly detection techniques that address inherent data challenges.
Improved safety monitoring for CPS is crucial for preventing catastrophic failures in critical infrastructure, autonomous systems, and industrial operations, directly impacting economic stability and public safety.
The proposed 'Uncertainty-Guided Label Rebalancing' technique offers a novel way to address extreme class imbalance in safety data for time-series CPS telemetry, potentially leading to more reliable and generalizable AI safety predictors.
- · Cyber-Physical Systems operators
- · AI safety researchers
- · Industries deploying autonomous systems
- · Software and AI development firms
- · Systems with high failure rates
- · Current standard rebalancing techniques
- · Organizations relying on simple anomaly detection
More accurate safety predictions in critical CPS environments.
Increased adoption of AI and automation in high-stakes CPS applications due to improved safety assurances.
Reduced operational costs and insurance premiums for industries utilizing advanced CPS with enhanced safety capabilities.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG