Quantum machine learning provides a fundamentally novel and promising approach to analyzing data. However, many data sets are too complex for currently available quantum computers. Consequently, quantum machine learning applications conventionally resort to dimensionality reduction algorithms, e.g., auto-encoders, before passing data through the quantum models. We show that using a classical auto-encoder as an independent preprocessing step can significantly decrease the classification performance of a quantum machine learning algorithm. To ameliorate this issue, we design an architecture that unifies the preprocessing and quantum classification algorithms into a single trainable model: the guided quantum compression model. The utility of this model is demonstrated by using it to identify the Higgs boson in proton-proton collisions at the LHC, where the conventional approach proves ineffective. Conversely, the guided quantum compression model excels at solving this classification problem, achieving a good accuracy. Additionally, the model developed herein shows better performance compared to the classical benchmark when using only low-level kinematic features.

Nature Machine Intelligence Dear Editors, We are hereby submitting the paper'AXXX' to Nature Machine Intelligence as we believe that the content fits the target audience of this Journal and the novelty criteria you require. To our knowledge the present study is the first demonstration of the application of graph neural networks for jet tagging on FPGAs for inference time within O(100) ns. Using the HLS4ML library combined with quantization-aware training and efficient FPGA implementations, we show that O(100) ns inference of complex architectures like graph convolutional neural networks, garnet and interaction networks is feasible at low resource-cost. Our target application is the real-time processing of Large Hadron Collider (LHC) data. However, we believe that the proposed solution could fit other problems related to low latency data selection beyond the LHC. The conditions at the LHC are unique and at the extreme end of the inference-on-the-edge spectrum.

The detection of out-of-distribution data points is a common task in particle physics. It is used for monitoring complex particle detectors or for identifying rare and unexpected events that may be indicative of new phenomena or physics beyond the Standard Model. Recent advances in Machine Learning for anomaly detection have encouraged the utilization of such techniques on particle physics problems. This review article provides an overview of the state-of-the-art techniques for anomaly detection in particle physics using machine learning. We discuss the challenges associated with anomaly detection in large and complex data sets, such as those produced by high-energy particle colliders, and highlight some of the successful applications of anomaly detection in particle physics experiments.