Accelerating the search for antimicrobial peptides with Google Antigravity
deepmind.google·2h ago
AI Research
arXiv CS.AI·3d ago·5 min read
Local Diagnostics of Learned Event Representations for Anomaly Detection at the HL-LHC
TL;DR
A novel approach for anomaly detection at the High-Luminosity Large Hadron Collider (HL-LHC) utilizes supervised and self-supervised neural embeddings to identify rare collision events. The method improves signal efficiency while providing a local diagnostic pipeline for analyzing flagged events.
✦ Why It Matters
Implement supervised learning techniques for anomaly detection in your own data analysis projects to enhance signal identification.
Key Takeaways
- Utilize supervised learning for better anomaly detection outcomes.
- Explore neural embeddings to analyze high-dimensional data efficiently.
- Implement a local diagnostic pipeline for event analysis.
- Rank observables to improve follow-up search strategies.
How It Works
The approach uses an autoencoder to learn low-dimensional representations of collision events, where the reconstruction error serves as an anomaly score. By fitting a Gaussian Mixture Model to the learned latent space, the method profiles flagged events and ranks observables based on their ability to distinguish between standard and anomalous events.