Rationale

As modern astrophysical surveys deliver an unprecedented amount of data, from the imaging of hundreds of millions of distant galaxies to the mapping of cosmic radiation fields at ultra-high resolution, conventional data analysis methods are reaching their limits in both computational complexity and optimality. Deep Learning has rapidly been adopted by the astronomical community as a promising way of exploiting these forthcoming big-data datasets and of extracting the physical principles that underlie these complex observations. This has led to an unprecedented exponential growth of publications combining Machine Learning and astrophysics. Yet, many of these works remain at an exploratory level and have not been translated into real scientific breakthroughs.

Following a successful initial iteration of this workshop at ICML 2022, our continued goal for this workshop series is to bring together Machine Learning researchers and domain experts in the field of Astrophysics to discuss the key open issues which hamper the use of Deep Learning for scientific discovery, and to present high-quality and cutting-edge work at the intersection between machine learning and astrophysics.

An important aspect to the success of Machine Learning in Astrophysics is to create a two-way interdisciplinary dialog in which concrete data-analysis challenges can spur the development of dedicated Machine Learning tools, which this workshop aims to facilitate. We expect this workshop to appeal to ICML audiences as an opportunity to connect their research interests to concrete and outstanding scientific challenges.

We welcome in particular submissions that target or report on the following non-exhaustive list of problems:

• Efficient high-dimensional inference
• Robustness to covariate shifts and model misspecification
• Anomaly and outlier detection, search for rare signals with ML
• Methods for accurate uncertainty quantification
• Methods for improving interpretability of models
• (Astro)-physics informed models, symmetry and equivariance-preserving models
• Methods of emulation / acceleration of simulation models
• Benchmarking and deployment of ML models for large-scale data analysis

We encourage both submissions on these topics with an astrophysics focus, as well as more methodologically oriented works with potential applications in the physical sciences.

Program

Invited Speakers and Panelists

Workshop Schedule

All times are in Hawaii Time. Please visit the ICML Workshop Page for live schedule (requires registration).

9:00:9:05	Introduction	Welcome and Introduction of the Workshop
9:05-9:35	Keynote	Chelsea Finn: Detecting and Adapting to Distribution Shift
9:35-9:50	Spotlight	Vidhi Ramesh: Shared Stochastic Gaussian process Decoders: A Probabilistic Generative model for Quasar Spectra
9:50-10:05	Spotlight	Yitian Sun: Disentangling gamma-ray observations of the Galactic Center using differentiable probabilistic programming
10:05-10:30	Break	Morning Coffee Break
10:30-11:00	Keynote	Anna Scaife: Foundation Models for Radio Astronomy
11:00-11:15	Spotlight	Guillermo Cabrera-Vives: Positional Encodings for Light Curve Transformers: Playing with Positions and Attention
11:15-11:30	Spotlight	Alice Desmons: Detecting Tidal Features using Self-Supervised Learning
11:30-11:45	Spotlight	Jonas Wildberger: Flow Matching for Scalable Simulation-Based Inference
11:45-12:00	Spotlight	Eve Campeau-Poirier: Time Delay Cosmography with a Neural Ratio Estimator
12:00-13:00	Break	Lunch Break
13:00-13:30	Keynote	Dmitry Duev: Astrophysics Meets MLOps
13:30-13:45	Spotlight	Carolina Cuesta: Diffusion generative modeling for galaxy surveys: emulating clustering for inference at the field level
13:45-14:00	Spotlight	Adrian Bayer: Field-Level Inference with Microcanonical Langevin Monte Carlo
14:00-14:15	Spotlight	Matt Sampson: Spotting Hallucinations in Inverse Problems with Data-Driven Priors
14:15-14:45	Keynote	Ross Taylor: Teaching LLMs to Reason
14:45-16:00	Poster Session	Main Poster Session
16:00-17:00	Panel Discussion	How will new technologies such as foundation models/generative models/LLMs change the way we do scientific discoveries? Panelists: Megan Andsell, Yashar Hezaveh, David W. Hogg, Peter Melchior, Irina Rish, Yuan-Sen Ting

Accepted Contributions

Cosmological Data Compression and Inference with Self-Supervised Machine Learning	Akhmetzhanova, Aizhan*; Mishra-Sharma, Siddharth; Dvorkin, Cora
Bayesian Uncertainty Quantification in High-dimensional Stellar Magnetic Field Models	Andersson, Jennifer R*; Kochukhov, Oleg ; Zhao, Zheng; Sjölund, Jens
Field-Level Inference with Microcanonical Langevin Monte Carlo	Bayer, Adrian*; Seljak, Uros; Modi, Chirag
Graph Representation of the Magnetic Field Topology in High-Fidelity Plasma Simulations for Machine Learning Applications	Bouri, Ioanna*; Franssila, Fanni; Alho, Markku; Cozzani, Giulia; Zaitsev, Ivan; Palmroth, Minna; Roos, Teemu
Domain Adaptation via Minimax Entropy for Real/Bogus Classification of Astronomical Alerts	Cabrera-Vives, Guillermo*; Bolívar, César Andrés; Förster, Francisco; Muñoz Arancibia, Alejandra M.; Pérez-Carrasco, Manuel; reyes, esteban dirk
Time Delay Cosmography with a Neural Ratio Estimator	Campeau-Poirier, Ève*; Perreault-Levasseur, Laurence; Coogan, Adam; Hezaveh, Yashar
A Comparative Study on Generative Models for High Resolution Solar Observation Imaging	Cherti, Mehdi*; Czernik, Alexander; Kesselheim, Stefan; Effenberger, Frederic; Jitsev, Jenia
Harnessing the Power of Adversarial Prompting and Large Language Models for Robust Hypothesis Generation in Astronomy	Ciuca, Ioana*; Ting, Yuan-Sen; Kruk, Sandor; Iyer, Kartheik
Diffusion generative modeling for galaxy surveys: emulating clustering for inference at the field level	Cuesta, Carolina*; Mishra-Sharma, Siddharth
Multiscale Flow for Robust and Optimal Cosmological Analysis	Dai, Biwei*; Seljak, Uros
Detecting Tidal Features using Self-Supervised Learning	Desmons, Alice*; Brough, Sarah; Lanusse, Francois
Multi-fidelity Emulator for Cosmological Large Scale 21 cm Lightcone Images: a Few-shot Transfer Learning Approach with GAN	Diao, Kangning*; Mao, Yi
SimBIG: Galaxy Clustering beyond the Power Spectrum	Hahn, ChangHoon*; Lemos, Pablo; Régaldo-Saint Blancard, Bruno; Parker, Liam H; Eickenberg, Michael; Ho, Shirley; Hou, Jiamin; Massara, Elena ; Modi, Chirag; Moradinezhad Dizgah, Azadeh; Spergel, David
Cosmology with Galaxy Photometry Alone	Hahn, ChangHoon*; Melchior, Peter M; Villaescusa-Navarro, Francisco; Teyssier, Romain
Shared Stochastic Gaussian process Decoders: A Probabilistic Generative model for Quasar Spectra	Lalchand, Vidhi *; Eilers, Anna-Christina
Closing the stellar labels gap: An unsupervised, generative model for Gaia BP/RP spectra	Laroche, Alexander L*; Speagle, Joshua S
Towards Unbiased Gravitational-Wave Parameter Estimation using Score-Based Likelihood Characterization	Legin, Ronan*; Wong, Kaze; Isi, Maximiliano; Adam, Alexandre; Perreault-Levasseur, Laurence; Hezaveh, Yashar
SimBIG: Field-level Simulation-based Inference of Large-scale Structure	Lemos, Pablo*; Parker, Liam H; Hahn, ChangHoon; Ho, Shirley; Eickenberg, Michael; Hou, Jiamin; Massara, Elena ; Modi, Chirag; Moradinezhad Dizgah, Azadeh; Régaldo-Saint Blancard, Bruno; Spergel, David
Using Multiple Vector Channels Improves $E(n)$-Equivariant Graph Neural Networks	Levy, Daniel*; Kaba, Sékou-Oumar; Gonzales, Carmelo; Miret, Santiago; Ravanbakhsh, Siamak
Population-Level Inference for Galaxy Properties from Broadband Photometry	Li, Jiaxuan*; Melchior, Peter M; Hahn, ChangHoon; Huang, Song
A Hierarchy of Normalizing Flows for Modelling the Galaxy-Halo Relationship	Lovell, Christopher C*; Hassan, Sultan; Villaescusa-Navarro, Francisco; Genel, Shy; Hahn, ChangHoon; Angles-Alcazar, Daniel; Kwon, James; de Santi, Natali; Iyer, Kartheik; Fabbian, Giulio; Bryan, Greg
PPDONet: Deep Operator Networks for Fast Prediction of Steady-State Solutions in Disk-Planet Systems	Mao, Shunyuan*; Dong, Ruobing; Lu, Lu; Yi, Kwang Moo; Wang, Sifan; Perdikaris, Paris
Positional Encodings for Light Curve Transformers: Playing with Positions and Attention	Moreno-Cartagena, Daniel A; Cabrera-Vives, Guillermo*; Protopapas, Pavlos; Donoso, Cristobal R; Pérez-Carrasco, Manuel Ignacio; Cádiz-Leyton, Martina A
Neural Astrophysical Wind Models	Nguyen, Dustin*
FLORAH: A generative model for halo assembly histories	Nguyen, Tri*; Modi, Chirag; Somerville, Rachel; Yung, Aaron
Multi-Class Deep SVDD: Anomaly Detection Approach in Astronomy with Distinct Inlier Categories	Pérez-Carrasco, Manuel Ignacio*; Cabrera-Vives, Guillermo; Hernandez-García, Lorena; Förster, Francisco; Sanchez-Saez, Paula; Muñoz-Arancibia, Alejandra; Astorga, Nicolás; Bauer, Franz; Bayo, Amelia; Cádiz-Leyton, Martina A; Catelan, Márcio; Estevez, Pablo
BTSbot: A Multi-input Convolutional Neural Network to Automate and Expedite Bright Transient Identification for the Zwicky Transient Facility	Rehemtulla, Nabeel*; Miller, Adam; Coughlin, Michael; Jegou Du Laz, Theophile
Toward a Spectral Foundation Model: An Attention-Based Approach with Domain-Inspired Fine-Tuning and Wavelength Parameterization	Różański, Tomasz; Ting, Yuan-Sen*; Jablonska, Maja
Spotting Hallucinations in Inverse Problems with Data-Driven Priors	Sampson, Matt L*; Melchior, Peter M
Evaluating Summary Statistics with Mutual Information for Cosmological Inference	Sui, Ce*; Zhao, Xiaosheng; Jing, Tao; Mao, Yi
Disentangling gamma-ray observations of the Galactic Center using differentiable probabilistic programming	Sun, Yitian*; Mishra-Sharma, Siddharth; Slatyer, Tracy R; Wu, Yuqing
Weisfeiler-Lehman Graph Kernel Method: A New Approach to Weak Chemical Tagging	Ting, Yuan-Sen*; Sharma, Bhavesh
A Novel Application of Conditional Normalizing Flows: Stellar Age Inference with Gyrochronology	Van-Lane, Phil*; Speagle, Joshua S; Douglas, Stephanie
Flow Matching for Scalable Simulation-Based Inference	Wildberger, Jonas Bernhard*; Dax, Maximilian; Buchholz, Simon; Green, Stephen R; Macke , Jakob; Schölkopf, Bernhard
Learning the galaxy-environment connection with graph neural networks	Wu, John F*; Jespersen, Christian
Diffusion Models for Probabilistic Deconvolution of Galaxy Images	Xue, Zhiwei; Li, Yuhang; Patel, Yash P*; Regier, Jeffrey
A cross-modal adversarial learning method for estimating photometric redshift of quasars	Zhang, Chen*; Zhang, Yanxia; Jiang, Bin; Qu, Meixia; Wang, Wenyu
nbi: the Astronomer's Package for Neural Posterior Estimation	Zhang, Keming*; Bloom, Joshua; Hernitschek, Nina
Stellar Spectra Fitting with Amortized Neural Posterior Estimation and nbi	Zhang, Keming*; Jayasinghe, Tharindu; Bloom, Joshua
3D ScatterNet: Inference from 21 cm Light-cones	Zhao, Xiaosheng*; Zuo, Shifan; Mao, Yi

---

Call for Abstracts

We invite all contributions in connection with the theme of the workshop as described here.

An important selection criterion will be the novelty of the work - novel methodologies and novel applications. Applications of standard and established deep learning techniques to a new astrophysical data set are not considered as novel applications in this context.

Original contributions and early research works are encouraged. Contributions presenting recently published or currently under review results are also welcome. The workshop will not have formal proceedings, but accepted submissions will be linked on the workshop webpage.

Submissions, in the form of extended abstracts, need to adhere to the ICML 2023 format (LaTeX style files), be anonymized, and be no longer than 4 pages (excluding references). After double-blind review, a limited set of submissions will be selected for contributed talks, and a wider set of submissions will be selected for poster presentations.

Please submit your anonymized extended abstract through CMT at https://cmt3.research.microsoft.com/ML4Astro2023 before May 19th May 25th, 23:59 AOE.

---

Volunteering to serve as a Reviewer

Our goal is to ensure that all extended abstracts will receive at least two independent reviews, in a double-blind process. As we aim for high quality and constructive reviews, we do not want to ask volunteers to review many papers, which translates into needing a large pool of volunteers.

As a result we are always looking for volunteers to help us review workshop submissions. If you are interested in serving as a reviewer, please let us know through this form before May 25th. Depending on the reviewing needs we may then contact you with further details and to confirm your availability.

---

Logistics and FAQs

ICML 2023 is currently planned as an in-person event. As such, this workshop is currently assuming a hybrid format, with physical poster sessions and in-person speakers, but with support for virtual elements to facilitate participation of people unable to travel. We encourage all interested participants (regardless of their ability to physically travel to ICML) to submit an extended abstracts.

Registration for ICML workshops is handled through the main ICML conference registration here.

Inquiries regarding the workshop can be directed to icml2023ml4astro@gmail.com

Important Dates

All dates are in AOE (Anywhere on Earth).

• **Submission deadline</b>: May 19th May 25th (23:59 AOE)
• **Author Notification</b>: June 16th 19th
• **Camera-ready paper deadline</b>: July 10th
• **Camera-ready poster deadline</b>: July 21st (see instructions below)
• **Workshop date</b>: Saturday July 29th, 9am to 5pm local time in Honolulu, Hawaii (UTC-10)

Instructions for Posters

To prepare your poster, please use the following dimensions:

• Recommended size: 24”W x 36”H, 61 x 90cm (A1 portrait)
• Maximum size: 48”W x 36”H, 122 x 90 cm (A0 landscape)
• Please use lightweight paper for printing, poster boards may not be available for workshops, but you will be able to tape your poster to the wall using provided tape.

---

SOC

Scientific Organizing Committee for the ICML 2023 Machine Learning for Astrophysics workshop: