gamma_flow: Denoise, classify and disentangle spectral data!
summary
gamma_flow is an open-source Python package that performs real‑time analysis of gamma spectra by using a supervised non‑negative matrix factorization to map 256‑channel spectra into a K‑dimensional latent space (one axis per isotope), enabling denoising by reconstruction, multi‑label classification with fractional shares, and cosine similarity outlier detection. It avoids neural nets for a lightweight workflow, reports classification accuracy above 90% on tested isotopes, and fails predictably when a class is omitted from training.
AI-written summary · about
chapters (13)
- 0:00Introduction and team
- 0:43Gamma spectra riddle
- 1:27Use case and challenges
- 3:13Goals for automation
- 4:16Introducing gamma flow
- 5:40Dimensionality reduction concept
- 7:52Model design and NMF
- 10:03Classification in latent space
- 11:30Testing with real data
- 12:57Denoising approach
- 14:01Outlier detection feature
- 15:48Generalization and outlook
- 16:31Q&A
related talks
47mWaveGlow: Generative Modeling for Audio Synthesis
32mpyGAM: balancing interpretability and predictive power using...
24mAn Open Channel Identifier using GNU Radio
29mTowards GridFM: GENCO – a Unified Neural Solver for Steady-stat...
37mThe solution of inverse problems
49mGraph Based Processing of Big Images
31mGridFM - Towards Foundation Models for the Electric Grid
20mAI in the physical world: from observation to discovery