IAA-CSIC Severo Ochoa School on Artificial Intelligence and Machine Learning in Astronomy
THE SCHOOL
Overview & Scientific Rationale
Modern astronomy is entering a data-intensive era in which Artificial Intelligence and Machine Learning are becoming essential scientific tools. From interferometric image reconstruction to source classification and cosmological inference, AI methods are transforming how astronomical discoveries are made.
The Instituto de Astrofísica de Andalucía – Severo Ochoa (IAA-SO) School on Artificial Intelligence and Machine Learning in Astronomy 2026 will provide graduate students, postdoctoral researchers, and early-career scientists with practical training in state-of-the-art AI techniques applied to real astronomical datasets from world-leading facilities.
Training Modules
The school is organized around five major technical pillars, each led by an expert in the field:
Image Reconstruction: Optical/infrared interferometry deconvolution utilizing Deep Image Priors Neural Networks (DIP-NN).
Radio Surveys: Data augmentation and automated morphological classification of radio sources using CNNs and Vision Transformers (ViT).
Large-Scale Structure: Graph Neural Networks (GNNs) for classification and regression tasks on dark matter halo simulations.
Gamma-Ray Astronomy: Deep learning-based image and waveform reconstruction for Imaging Atmospheric Cherenkov Telescopes.
Unsupervised Learning: High-dimensional embeddings (t-SNE, UMAP, EVoC) and clustering algorithms applied to chemical tagging of star clusters.
Who should attend?
The school is designed for:
- MSc and PhD students in astronomy, physics, data science, or related disciplines
- Postdoctoral researchers
- Early-career scientists interested in AI applications in astronomy
- Researchers seeking practical experience with modern machine learning workflows
Prerequisites
Participants are expected to have:
- Basic Python programming experience.
- Familiarity with scientific computing tools (NumPy, Jupyter notebooks).
- Basic astronomy knowledge
Computing Environment
Tutorials will be delivered through Python notebooks. Software installation instructions and datasets will be distributed before the school. Participants should bring a laptop capable of running Python scientific workflows.
Learning outcomes
Participants will:
- Build and train modern neural network architectures
- Work directly with astronomical FITS datasets and simulations
- Develop practical experience with CNNs, Vision Transformers, Graph Neural Networks and unsupervised learning methods
- Understand strengths and limitations of AI approaches in astronomy
- Gain reproducible workflows applicable to their own research
Speakers & Lecturers
We are pleased to host an international team of lecturers specializing in different subsets of astronomical machine learning:
- Dr. Joel Sánchez Bermúdez Instituto de Astronomía (IA-UNAM), Mexico
- Dr. Andrea DeMarco Institute of Space Sciences and Astronomy (ISSA), University of Malta, Malta
- Dr. Farida Farsian Italian National Institute for Astrophysics (INAF), Osservatorio Astrofisico di Catania (OACT), Italy
- Dr. Tjark Miener University of Geneva, Switzerland / IAA Granada, Spain
- Dr. Rafael Garcia-Dias, King’s College London, United Kingdom
SCIENTIFIC ORGANIZING COMMITTEE
- Dr. Joel Sánchez Bermúdez (IA-UNAM, Mexico) — Chair
- Dr. Javier Moldón (IAA-CSIC, Spain) — Co-chair
- Dr. Cristóbal Bordiú (IAA-CSIC, Spain)
- Dr. Laura Darriba (IAA-CSIC, Spain)
- Dr. Rubén López-Coto (IAA-CSIC, Spain)
- Dr. Ginés Martínez Solaeche (IAA-CSIC, Spain)
LOCAL ORGANIZING COMMITTEE
- Dr. Laura Darriba (IAA-CSIC)
- Dr. Javier Moldón (IAA-CSIC)
- Dr. Cristóbal Bordiú (IAA-CSIC)