The Machine Learning for Transient Science (MALTS) Topical Team within DESC is committed to the advancement of statistical tools for studying the variable sky. With the advent of the Vera Rubin Observatory's Legacy Survey for Space and Time starting in 2024, automated approaches to data anlysis will become increasingly valuable. At our tri-weekly meetings, we aim to discuss new papers in the field, offer tutorials for popular ML methods, and highlight current ML-oriented projects within DESC.
- October 27th, 2022: First meeting, general structure and meeting goals.
- November 17th, 2022: Survey discussion, Graph Neural Networks and CNNs for DIA.
- December 1st, 2022: Introduction to autodifferentiation with JAX, Normalizing Flows with PZFlow.
- December 15th, 2022: Hack session on Normalizing Flows.
- January 25th, 2023: Bayesian inference with Rubin Supernovae and Galaxies.
To request a new machine learning tutorial, please click here. Have you seen a paper you'd like to discuss? Add it as an issue here (template borrowed from the awesome CosmoStat Laboratory)!
If you have a question about the group, please reach out to co-leads Alex Gagliano ([email protected]) or Benjamin Remy ([email protected]).
Note: The logo for this group was generated using midjourney, a text-to-image program based on diffusion models. For an overview of how diffusion works for image generation, check out this brief introduction.