Center for Decoding the Universe Annual Conference

The Center for Decoding the Universe brings together researchers across scientific disciplines to answer the biggest questions about our Universe by leveraging complex data with the most advanced computational methods. 

Conference Motivation

From anomaly detection in massive data sets and foundation models that compactly describe multi-modal data, to simulation-based inference that bridges the gap between the observed and simulated Universe, new methodologies will enable new insights from large and complex datasets. These new approaches are poised to impact the next big thing in astrophysics: data from the Vera C Rubin Observatory’s Legacy Survey of Space and Time (LSST). This meeting will gather leading researchers in astrophysics, AI/ML, data science, and statistics, and will identify new opportunities for inference and data-driven discovery with the imminent LSST data.

Conference Format

With time devoted to small group discussion and half a day of unconference, the two-day conference is designed to promote discussion about the hard questions and methodological innovations that will be especially relevant in the upcoming large astronomical data era. Three sessions that combine an astronomical subfield with a data science/AI methodology will focus on:

1. Anomaly detection with applications to variable and transient science
2. Astronomical foundation models and applications to galaxy evolution
3. Inference, including simulation-based inference, with applications to cosmology

Each session will be introduced by three speakers who will together (1) outline the current state of the astronomical field, emphasizing major open questions that LSST data could help address, (2) provide a broad overview of the methodology, covering recent advances while highlighting existing limitations that may hinder its full potential when applied to astronomical data, and (3) pose challenging questions about the methodology’s applicability and effectiveness in addressing open astronomical questions compared to traditional approaches.

Following the introductory session, participants will break into small group discussions, followed by group-wide report-backs. 

The half-day unconference on the second day is designed to foster further discussion and collaborative problem-solving, allowing small groups to explore potential answers and solutions to the questions raised in the three sessions.

We welcome contributed talks and posters on any aspect of AI/ML, data science, statistics, and/or their intersection with the physics of the universe. We invite submissions focused on methodologies, science, or how science is done, including the role of agentic AI in analysis and discovery.

Conference Schedule Outline

Call for Short Talks and Posters

We invite you to submit an abstract to present a 5-10 minute short talk and/or poster at the upcoming 2025 Center for Decoding the Universe Annual Conference. Abstracts for short talks and/or posters should showcase any aspect of your research relevant to data‑driven discovery at scale from anomaly detection in time‑domain surveys and foundation models for multi‑modal data, to simulation‑based inference that links theory and observation. Talks should be accessible to researchers spanning astrophysics, AI/ML, data science, and statistics. Work in progress is welcome.

Deadline: May 12, 2025 by 5:00 PM PT