<ul><li>Complex events (CEs) are important in CPS-IoT applications for high-level decision-making.</li><li>Most existing models focus on short-span perception tasks and lack the long-term reasoning required for CE detection.</li><li>This case study explores different approaches for CE detection in CPS-IoT, including leveraging large language models, employing neural architectures, and adopting a neurosymbolic approach.</li><li>The results show that the state-space model called Mamba outperforms other methods in accuracy and generalization to longer, unseen sensor traces.</li></ul>

Toward Foundation Models for Online Complex Event Detection in CPS-IoT: A Case Study

Discover more