Enhancing anomaly detection with topology-aware autoencoders
Vishal S. Ngairangbam, Błażej Rozwoda, Kazuki Sakurai, Michael Spannowsky · 2025
Read Enhancing anomaly detection with topology-aware autoencoders by Vishal S. Ngairangbam, Błażej Rozwoda, Kazuki Sakurai, Michael Spannowsky (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-10163.txt.
Description
Vishal Ngairangbam and co-authors sharpen the hunt for new physics in 2025 with autoencoders that respect shape. Such networks learn to rebuild normal collision data, then flag events they rebuild badly as odd. Their weak point is the shape of the hidden space where data is squeezed, which is usually a plain flat space.
The team instead squeezes data onto curved shapes, such as spheres and paired spheres, picked to mirror energy and momentum rules. On mock decays of top quarks, the shaped versions split odd events from normal ones more cleanly, they report. False rebuild errors fell as the hidden shape kept the data's true form. Shape-aware squeezing offers a stronger blind search tool for the next collider runs.
Catalog record
| CREATOR | Vishal S. Ngairangbam, Błażej Rozwoda, Kazuki Sakurai, Michael Spannowsky |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 44 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-10163