Fast Machine Learning for Science Conference 2026

Real-time Quantum Inspired Algorithms for Anomaly Detection in Collider Triggers

Not scheduled

12m

Presentation Contributed Talks

Speaker

Sagar Addepalli (SLAC National Accelerator Laboratory (US))

Description

Real-time Quantum Inspired Algorithms for Anomaly Detection in Collider Triggers
Anomaly detection algorithms deployed at the edge of particle collider experiments offer a model-agnostic approach to search for signs of new physics, complementing typical topology- or energy-driven selections. Tensor networks are a class of quantum-inspired machine learning models that represent information in low-rank tensor factorizations rather than the dense weight matrices of fully connected neural networks. In this talk, we show that such tensor networks, specifically Spaced Matrix Product Operators (SMPOs), are capable of performing real-time anomaly detection at colliders, achieving high discrimination on benchmark new-physics signals while meeting the sub-microsecond-scale latency and resource budgets of current trigger and readout systems. The fully linear structure of SMPOs admits systematic tensor factorization, adding flexibility for resource-aware deployment. We further demonstrate a standardized approach to pruning sparse structure of SMPOs, substantially reducing model parameters and operations with minimal loss in physics performance. As one of the first feasibility studies of tensor networks for unsupervised machine learning targeting resource-constrained collider trigger environments, this work lays the foundation for a new class of malleable algorithms well suited to real-time deployment.