Loading...
Organizing Committee
University of Massachusetts Dartmouth
Caltech
University of Toronto
Brown University
The University of British Columbia
University of Rhode Island
University of Toronto
University of Massachusetts, Dartmouth
Abstract
The aim of this workshop is to bring together participants from computational mathematics and gravitational wave astronomy to tackle computational challenges in leveraging data-driven methods in key areas of gravitational wave data analysis in order to maximize the science output of the ongoing and upcoming observations. The areas of focus will be: (i) noise classification and detection, (ii) waveform modeling and uncertainty quantification, and (iii) source parameter and astrophysical population Bayesian inference.
The participants will develop and apply new mathematical and computational techniques including: (i) neural network classifiers for distinguishing signals from instrumental noise, (ii) generative machine learning models for simulating realizations of non-Gaussian and non-stationary stochastic processes, (iii) surrogate models including uncertainty quantification, (iv) stochastic sampling, neural posterior estimation leveraging deep neural networks with normalizing flows or diffusion models, and (v) hierarchical Bayesian inference with non-parametric models such as Gaussian processes and simulation-based / approximate Bayesian approaches.
The participants will develop and apply new mathematical and computational techniques including: (i) neural network classifiers for distinguishing signals from instrumental noise, (ii) generative machine learning models for simulating realizations of non-Gaussian and non-stationary stochastic processes, (iii) surrogate models including uncertainty quantification, (iv) stochastic sampling, neural posterior estimation leveraging deep neural networks with normalizing flows or diffusion models, and (v) hierarchical Bayesian inference with non-parametric models such as Gaussian processes and simulation-based / approximate Bayesian approaches.
Image credit: Max Planck Institute for Gravitational Physics, Simulating eXtreme Spacetimes Collaboration and Airborne Hydro Mapping GmbH
Loading...
Loading participant list in background...
Workshop Schedule
Monday, June 02, 2025
8:20 - 8:40 AM EDT
Check In
11th Floor Collaborative Space
8:40 - 9:00 AM EDT
Welcome
11th Floor Lecture Hall
Brendan Hassett, ICERM/Brown University
9:00 - 9:30 AM EDT
Addressing the challenges of characterizing gravitational-wave interferometers with machine learning
11th Floor Lecture Hall
Speaker
Derek Davis, California Institute of Technology
Session Chair
Melissa Lopez, Utrecht University
Abstract
Future observing runs with ground-based gravitational-wave interferometers promise to rapidly increase the detection rate of signals and the breadth of science we can probe with gravitational waves. However, these discoveries are often hampered by the realistic features of detector data that violate many of the core assumptions of our current data analysis methods. To address these problems, numerous techniques are employed to identify and mitigate problems in the detectors that may impact the detection or analysis of gravitational-wave signals. As the event rate grows, these procedures will also need to evolve to quickly and generically address the wide variety of problems that may impact each new signal. In this talk, I will detail the current state-of-the-art methods to characterize the performance of gravitational-wave detectors, address common data issues that impact the analysis of transient gravitational-wave signals, and explore how the field can adapt to the big data challenges on the horizon.
9:45 - 10:15 AM EDT
AI-Powered Gravitational-Wave Detection: Bridging Traditional Pipelines and Neural Networks
11th Floor Lecture Hall
Speaker
Surabhi Sachdev, Georgia Institute of Technology
Session Chair
Melissa Lopez, Utrecht University
Abstract
Gravitational-wave astronomy has entered a new phase of discovery, with detectors recording hundreds of compact binary coalescences and preparing for even more sensitive observing runs. As the scale and complexity of the data continue to grow, so do the demands on search pipelines to be fast, efficient, and robust. Recent advances in artificial intelligence, particularly deep learning and neural networks, present exciting opportunities to enhance gravitational-wave detection capabilities. In this talk, I will highlight emerging machine learning approaches, discuss their current limitations, and explore what it would take for artificial intelligence methods to integrate seamlessly into production-level pipelines.
10:15 - 10:40 AM EDT
Coffee Break
11th Floor Conference Room
10:40 - 10:50 AM EDT
Poster Blitz
Lightning Talks - 11th Floor Collaborative Space
Speaker
Cecilia Fabbri, University of Nottingham
Speaker
Francesco Nobili, University of insubria
Speaker
Filippo Santoliquido, Gran Sasso Science Institute
Speaker
Rodrigo Tenorio, Università degli Studi di Milano-Bicocca
Speaker
Digvijay (Jay) Wadekar, Johns Hopkins University
Speaker
He Wang, University of Chinese Academy of Sciences
Session Chair
Vijay Varma, University of Massachusetts, Dartmouth
11:00 AM - 12:15 PM EDT
Neural network classifiers for distinguishing signals from instrumental noise
Tutorial - 11th Floor Lecture Hall
Speaker
Melissa Lopez, Utrecht University
Session Chair
Anushka Doke, University of Massachusetts Dartmouth
Abstract
In this tutorial, we perform a supervised multi-class classification task using triggers from matched filtering. Our neural network, built with Keras, distinguishes intermediate-mass binary black hole gravitational wave signals from various types of glitches. Through model training and evaluation, we explore how deep learning can separate astrophysical signals from instrumental noise in gravitational wave data.
12:15 - 1:30 PM EDT
Lunch/Free Time
1:30 - 2:00 PM EDT
Introduction to Discussion Topics
Program Overview - 11th Floor Lecture Hall
Session Chair
Derek Davis, California Institute of Technology
Session Chair
Melissa Lopez, Utrecht University
Session Chair
Surabhi Sachdev, Georgia Institute of Technology
2:00 - 3:30 PM EDT
Working Problem Session
Problem Session - Assigned Problem Session Locations
3:30 - 4:00 PM EDT
4:00 - 5:00 PM EDT
Open Problem Session Summaries
Problem Session - 11th Floor Lecture Hall
5:00 - 6:30 PM EDT
Reception
11th Floor Collaborative Space
Tuesday, June 03, 2025
9:00 - 9:40 AM EDT
An overview of gravitational wave surrogate modelling
11th Floor Lecture Hall
Speaker
Carl-Johan Haster, University of Nevada, Las Vegas
Session Chair
Vijay Varma, University of Massachusetts, Dartmouth
Abstract
I will present an introduction to the construction and use of surrogates in gravitational wave modelling. This will be an overview of modelling choices and conventions, acting as an entry point for the surrogate methods described in greater detail in the tutorials later in the day. I will also discuss a set of outstanding tasks that will need to be incorporated over the coming few years to prepare the surrogate models for the gravitational wave observations of the future.
9:50 - 10:15 AM EDT
Learning high-fidelity gravitational-wave models from numerical relativity data
Tutorial - 11th Floor Lecture Hall
Speaker
Scott Field, University of Massachusetts Dartmouth
Session Chair
Vijay Varma, University of Massachusetts, Dartmouth
Abstract
This tutorial will provide an introduction to publicly available code for building gravitational wave surrogate models. We will cover the construction of training datasets, temporal compression with linear dimensional reduction techniques, and fitting data over the model's parameter space. The methods covered in this tutorial will focus on "classical" techniques like singular value decomposition and interpolation-based methods. A second tutorial, covered by Lucy Thomas, will explore deep learning approaches to model building.
10:15 - 10:40 AM EDT
Coffee Break
11th Floor Conference Room
10:40 - 10:50 AM EDT
Poster Blitz
Lightning Talks - 11th Floor Lecture Hall
Speaker
Sofia Alvarez-Lopez, MIT
Speaker
Giulia Borghetto, University of Swansea
Speaker
Paola Carolina Delgado, CEICO - Institute of Physics of the Czech Academy of Sciences (FZU)
Speaker
Utkarsh Mali, University of Toronto
Speaker
Akash Maurya, International Centre for Theoretical Sciences
Speaker
Kaushik Paul, indian institute of technology madras
Speaker
Estuti Shukla, Pennsylvania State University
Session Chair
Vijay Varma, University of Massachusetts, Dartmouth
11:00 - 11:40 AM EDT
Public Code Packages to Visualise and Optimise Gravitational Wave Surrogate Models
Tutorial - 11th Floor Lecture Hall
Speaker
Lucy Thomas, Caltech
Session Chair
Michael Puerrer, University of Rhode Island
Abstract
In this tutorial, I will cover examples of how to use publicly-available code packages to evaluate and visualise gravitational wave surrogate models. I will also introduce machine-learning based approaches to building surrogate models, and show how to optimise their architecture in the building process.
11:50 AM - 12:15 PM EDT
Flexible parameter estimation and fast waveforms
Tutorial - 11th Floor Lecture Hall
Speaker
Kaze Wong, John Hopkins University
Session Chair
Michael Puerrer, University of Rhode Island
Abstract
I am going to talk about trade-offs in accelerating parameter estimation and some recent development in waveform models on accelerator
12:30 - 1:45 PM EDT
Lunch/Free Time
1:45 - 2:15 PM EDT
Idea Session/ Problem Proposals
Problem Session - 11th Floor Lecture Hall
Session Chair
Keefe Mitman, California Institute of Technology
Session Chair
Vijay Varma, University of Massachusetts, Dartmouth
2:15 - 3:30 PM EDT
Working Problem Session
Problem Session - Assigned Problem Session Locations
3:30 - 4:00 PM EDT
4:00 - 5:00 PM EDT
Open Problem Session Summaries
Problem Session - 11th Floor Lecture Hall
Wednesday, June 04, 2025
9:00 - 9:45 AM EDT
Gravitational wave inference and stochastic sampling
11th Floor Lecture Hall
Speaker
Colm Talbot, University Of Chicago
Session Chair
Katerina Chatziioannou, Caltech
9:45 - 10:15 AM EDT
Bilby
Tutorial - 11th Floor Lecture Hall
Speaker
Colm Talbot, University Of Chicago
Session Chair
Katerina Chatziioannou, Caltech
10:15 - 10:40 AM EDT
Coffee Break
11th Floor Conference Room
10:40 - 10:50 AM EDT
Poster Blitz
Lightning Talks - 11th Floor Lecture Hall
Speaker
Som Bishoyi, University of Massachusetts Dartmouth
Speaker
Matteo Boschini, University of Milano-Bicocca
Speaker
John Brennan, Maynooth University
Speaker
Chayan Chatterjee, Vanderbilt University
Speaker
Ashwin Girish, University of Rhode Island
Speaker
Sanaz Hami Hassan Kiyadeh, The University of Alabama
Speaker
Pranav Vinod, University of Massachusetts Dartmouth
Session Chair
Vijay Varma, University of Massachusetts, Dartmouth
11:00 - 11:45 AM EDT
Neural posterior estimation for gravitational-wave inference
11th Floor Lecture Hall
Speaker
Stephen Green, University of Nottingham
Session Chair
Michael Puerrer, University of Rhode Island
Abstract
I will describe how deep learning and simulation-based inference address gravitational-wave data analysis challenges, including high event rates and rapid electromagnetic follow-up. The approach uses simulated data to train neural networks, such as normalizing flows, to accurately represent posterior distributions. Once trained, these models enable extremely rapid inference—reducing analyses to seconds. I will highlight recent advances in population inference and binary neutron star parameter estimation, demonstrating the promise of these techniques for next-generation detectors.
11:45 AM - 12:15 PM EDT
Neural posterior estimation for gravitational-wave inference Continued
Tutorial - 11th Floor Lecture Hall
Speaker
Stephen Green, University of Nottingham
Session Chair
Michael Puerrer, University of Rhode Island
Abstract
I will describe how deep learning and simulation-based inference address gravitational-wave data analysis challenges, including high event rates and rapid electromagnetic follow-up. The approach uses simulated data to train neural networks, such as normalizing flows, to accurately represent posterior distributions. Once trained, these models enable extremely rapid inference—reducing analyses to seconds. I will highlight recent advances in population inference and binary neutron star parameter estimation, demonstrating the promise of these techniques for next-generation detectors.
12:30 - 1:45 PM EDT
Lunch/Free Time
1:45 - 2:15 PM EDT
Idea Session/ Problem Proposals
Problem Session - 11th Floor Lecture Hall
Hsin-Yu Chen, University Of Texas at Austin
Carl-Johan Haster, University of Nevada, Las Vegas
John Veitch, University of Glasgow
2:15 - 3:30 PM EDT
Working Problem Session
Problem Session - Assigned Problem Session Locations
3:30 - 4:00 PM EDT
4:00 - 5:00 PM EDT
Open Problem Session Summaries
Problem Session - 11th Floor Lecture Hall
Thursday, June 05, 2025
9:00 - 9:40 AM EDT
Introduction to gravitational-wave population inference
11th Floor Lecture Hall
Speaker
Aditya Vijaykumar, University of Toronto
Session Chair
Davide Gerosa, Università degli Studi di Milano-Bicocca
Abstract
In this talk, I will provide an overview of gravitational-wave (GW) population inference. I will begin by motivating the need to go beyond individual events and reconstructing the underlying source population. I will then introduce the framework of hierarchical Bayesian inference, with a focus on correcting for selection effects inherent in GW detectors. Finally, I will present the hierarchical likelihood, and present methods of computing it from the properties of individual events.
9:50 - 10:30 AM EDT
Tutorial on Hierarchical Bayesian Inference for CBC Populations
Tutorial - 11th Floor Lecture Hall
Speaker
Amanda Farah, University of Chicago
Session Chair
Davide Gerosa, Università degli Studi di Milano-Bicocca
Abstract
I will give a brief introduction to the framework of hierarchical Bayesian inference. I will then provide a simplified, hands-on example of how to implement such an analysis in the case of phenomenologically modeling the GW source population.
10:30 - 11:00 AM EDT
Coffee Break
11th Floor Conference Room
11:00 - 11:10 AM EDT
Poster Blitz
Lightning Talks - 11th Floor Lecture Hall
Speaker
Ritesh Bachhar, University of Rhode Island
Speaker
Samuel Clyne, University of Rhode Island
Speaker
Tousif Islam, University of California Santa Barbara
Speaker
Adhrit Ravichandran, University of Massachusetts, Dartmouth
Speaker
Anarya Ray, Northwestern University
Speaker
Javier Roulet, California Institute of Technology
Speaker
Kanchan Soni, Syracuse University
Session Chair
Vijay Varma, University of Massachusetts, Dartmouth
11:20 AM - 12:10 PM EDT
Machine learning in gravitational-wave population inference
Tutorial - 11th Floor Lecture Hall
Speaker
Matthew Mould, Massachusetts Institute of Technology
Session Chair
Aditya Vijaykumar, University of Toronto
Abstract
In this tutorial, I will demonstrate applications of machine learning for inference of source populations from gravitational-wave data. I will introduce several methods that can aid astrophysical interpretation of catalogs, including simulation-based priors, neural posterior estimation, and variational inference. The tutorial will then move to interactive notebooks for participants to explore these methods in practice on real data.
12:15 - 1:45 PM EDT
Lunch/Free Time
1:45 - 2:15 PM EDT
Idea Session/ Problem Proposals
Problem Session - 11th Floor Lecture Hall
Sylvia Biscoveanu, Northwestern University
Davide Gerosa, Università degli Studi di Milano-Bicocca
2:15 - 3:30 PM EDT
Working Problem Session
Problem Session - Assigned Problem Session Locations
3:30 - 4:00 PM EDT
4:00 - 5:00 PM EDT
Open Problem Session Summaries
Problem Session - 11th Floor Lecture Hall
Friday, June 06, 2025
9:00 - 9:30 AM EDT
Toward Online Deployment of Neural-Networks to Search and Characterize Compact Binary Mergers
11th Floor Lecture Hall
Speaker
Deep Chatterjee, MIT
Session Chair
Melissa Lopez, Utrecht University
Abstract
I will talk about a neural network (NN)-based search algorithm, Aframe, to detect binary black hole (BBH) mergers, and a NN-based parameter estimation algorithm, AMPLFI, to infer parameters of the detected signal. I will briefly talk about the algorithms and discuss the design choices for Aframe and AMPLFI to run in tandem on live LIGO-Virgo data to produce low-latency discovery alerts.
9:45 - 10:15 AM EDT
NextGen Tools for searches, analyses and detector characterization in LIGO
11th Floor Lecture Hall
Speaker
Sidd Soni, MIT
Session Chair
Melissa Lopez, Utrecht University
Abstract
In this work, we introduce a novel signal-versus-noise identification tool based on the YOLO (You Only Look Once) object detection framework. This tool provides scene identification capabilities and essential information regarding whether an observed transient is noise or signal. Additionally, it supplies detailed time-frequency coordinates in the form of pixel masks. The simultaneous identification of noise and signal, combined with precise pixel-level localization, represents a significant advancement in gravitational-wave data analysis. By rigorously validating our approach on real LIGO detector data, we demonstrate the potential of this method to improve the accuracy and reliability of gravitational-wave detections, thereby contributing meaningfully to the broader objectives of gravitational-wave astronomy.
10:15 - 10:40 AM EDT
Coffee Break
11th Floor Conference Room
10:45 - 11:25 AM EDT
GWSkyNet: an LVK machine learnig pipeline for gravitational wave candidate evaluation
11th Floor Lecture Hall
Speaker
Mervyn Chan, The University of British Columbia
Session Chair
Melissa Lopez, Utrecht University
Abstract
Multi-messenger observations of gravitational waves and their associated electromagnetic emission from compact binary coalescences offer unique insights into physics, astrophysics and cosmology. However, the presence of detector noise transients makes real-time searches for gravitational waves challenging. We developed GWSkyNet, an LVK machine learning pipeline capable of evaluating, and determining the origin of, gravitational wave candidates in a matter of seconds. We describe the successful application of GWSkyNet and its potential future development.
11:35 AM - 12:15 PM EDT
Building GSpyNetTree-II: a gravitational-wave signal vs glitch classifier for the fourth LIGO-Virgo-KAGRA observing run
11th Floor Lecture Hall
Speaker
Sofia Alvarez-Lopez, MIT
Session Chair
Melissa Lopez, Utrecht University
Abstract
LIGO and Virgo detector data contain frequent bursts of non-Gaussian transient noise, commonly known as glitches, which often mimic or overlap with gravitational wave (GW) signals. Given the higher candidate event rate in the current observing run (O4) compared to previous observing runs, the process of vetting GW candidates, including identifying glitches responsible for or overlapping with candidates, requires more automation than in previous runs. We present GSpyNetTree, the Gravity Spy Convolutional Neural Network Decision Tree: a multi-CNN multi-label classifier based on GravitySpy. Achieving an accuracy surpassing 96%, GSpyNetTree identifies glitches present in each GW detector at the time of a signal candidate. We have trained GSpyNetTree to be robust to a broad array of background noise, new sources of glitches, and the likely occurrence of overlapping glitches and GWs. We evaluate the performance of GSpyNetTree, report on its use in the O4 LIGO-Virgo event validation process, and suggest novel strategies to improve classifications.
12:15 - 1:30 PM EDT
Networking Lunch
Working Lunch - 11th Floor Collaborative Space
1:30 - 2:00 PM EDT
Idea Session/ Problem Proposals
Problem Session - 11th Floor Lecture Hall
Session Chair
Anushka Doke, University of Massachusetts Dartmouth
Session Chair
Melissa Lopez, Utrecht University
2:00 - 3:30 PM EDT
Working Problem Session
Problem Session - Assigned Problem Session Locations
3:30 - 4:00 PM EDT
Coffee Break
11th Floor Collaborative Space
4:00 - 5:00 PM EDT
Open Problem Session Summaries
Problem Session - 11th Floor Lecture Hall
Request Reimbursement
This section is for general purposes only and does not indicate that all attendees receive funding. Please refer to your personalized invitation to review your offer.
- ORCID iD
- As this program is funded by the National Science Foundation (NSF), ICERM is required to collect your ORCID iD if you are receiving funding to attend this program. Be sure to add your ORCID iD to your Cube profile as soon as possible to avoid delaying your reimbursement.
- Acceptable Costs
- 1 roundtrip between your home institute and ICERM
- Flights on U.S. or E.U. airlines – economy class to either Providence airport (PVD) or Boston airport (BOS)
- Ground Transportation to and from airports and ICERM.
- Unacceptable Costs
- Flights on non-U.S. or non-E.U. airlines
- Flights on U.K. airlines
- Seats in economy plus, business class, or first class
- Change ticket fees of any kind
- Multi-use bus passes
- Meals or incidentals
- Advance Approval Required
- Personal car travel to ICERM from outside New England
- Multiple-destination plane ticket; does not include layovers to reach ICERM
- Arriving or departing from ICERM more than a day before or day after the program
- Multiple trips to ICERM
- Rental car to/from ICERM
- Flights on a Swiss, Japanese, or Australian airline
- Arriving or departing from airport other than PVD/BOS or home institution’s local airport
- 2 one-way plane tickets to create a roundtrip (often purchased from Expedia, Orbitz, etc.)
- Travel Maximum Contributions
- New England: $350
- Other contiguous US: $850
- Asia & Oceania: $2,000
- All other locations: $1,500
- Note these rates were updated in Spring 2023 and supersede any prior invitation rates.
- Reimbursement Requests
Request Reimbursement with Cube
Refer to the back of your ID badge for more information. Checklists are available at the front desk and in the Reimbursement section of Cube.
- Reimbursement Tips
- Scanned original receipts are required for all expenses
- Airfare receipt must show full itinerary and payment
- ICERM does not offer per diem or meal reimbursement
- Allowable mileage is reimbursed at prevailing IRS Business Rate and trip documented via Google Maps PDF
- Keep all documentation until you receive your reimbursement!
- Reimbursement Timing
6–8 weeks after all documentation is sent to ICERM. All reimbursements are reviewed by multiple Brown University offices.
- Reimbursement Deadline
Submissions must be received within 30 days of ICERM departure to avoid applicable taxes. No submissions are accepted after six months.