Publications

Domain Adaptation for Simulation-Based Dark Matter Searches with Strong Gravitational Lensing

Published in NeurIPS: Machine Learning and the Physical Sciences Research Workshop, 2022

Paper detailing the continuation of my Google Summer of Code 2021 project now using more complex data sources and comparing Unsupervised Domain Adaptation between an EfficientNet and an Equivariant Neural Network architecture.

Recommended citation: Domain Adaptation for Simulation-Based Dark Matter Searches with Strong Gravitational Lensing. Stephon Alexander and Michael W. Toomey, Sergei Gleyzer, Pranath Reddy, Marcos Tidball. Neural Information Processing Systems (NeurIPS) Conference: Machine Learning and the Physical Sciences Research Workshop, 2022 https://ml4physicalsciences.github.io/2022/files/NeurIPS_ML4PS_2022_58.pdf

Similarity Search of Low Surface Brightness Galaxies in Large Astronomical Catalogs

Published in NeurIPS: LatinX in AI Research Workshop, 2022

Paper detailing lsh_astro, a tool that uses Locality-Sensitive Hashing with PySpark to perform an approximate similarity search of Low Surface Brightness Galaxies (LSBGs) in large astronomical catalogs. It allows for a quick and computationally efficient way for astronomers to find new LSBG candidates in large astronomical catalogs needing only one labeled LSBG.

Recommended citation: Similarity Search of Low Surface Brightness Galaxies in Large Astronomical Catalogs. Marcos Tidball, Cristina Furlanetto. Neural Information Processing Systems (NeurIPS) Conference: LatinX in AI (LXAI) Research Workshop, 2022. https://doi.org/10.52591/lxai202211282

(Undergraduate thesis, in Portuguese) Busca de galáxias de baixo brilho superficial por similaridade em grandes catálogos astronômicos

Published in Lume UFRGS, 2022

Undergraduate thesis detailing the creation of lsh_astro, a tool that uses Locality-Sensitive Hashing with PySpark to perform an approximate similarity search of Low Surface Brightness Galaxies (LSBGs) in large astronomical catalogs. It allows for a quick and computationally efficient way for astronomers to find new LSBG candidates in large astronomical catalogs needing only one labeled LSBGs.

Recommended citation: Busca de galáxias de baixo brilho superficial por similaridade em grandes catálogos astronômicos. Marcos Tidball, Cristina Furlanetto. Lume UFRGS, 2022. https://lume.ufrgs.br/handle/10183/255236

Domain Adaptation for Simulation-Based Dark Matter Searches Using Strong Gravitational Lensing

Published in arXiv preprint, 2021

Paper detailing my project at Google Summer of Code 2021 about an initial development of Unsupervised Domain Adaptation applied to gravitational lenses as well as an exploration of Equivariant Neural Networks for astronomical image classification.

Recommended citation: Domain Adaptation for Simulation-Based Dark Matter Searches Using Strong Gravitational Lensing. Stephon Alexander, Sergei Gleyzer, Pranath Reddy, Marcos Tidball, Michael W. Toomey. arXiv preprint, 2021 arxiv.org/abs/2112.12121

Deepfuse: Automatic Detection and Classification of Low Surface Brightness Galaxies with Convolutional Neural Networks

Published in Meetings of the Brazilian Astronomical Society, 2021

Poster detailing Deepfuse, a pipeline for the detection of Low Surface Brightness Galaxies with an image‑based DBSCAN algorithm and convolutional neural networks applied to large astronomical images fetched from an online database.

Recommended citation: Deepfuse: Automatic Detection and Classification of Low Surface Brightness Galaxies with Convolutional Neural Networks. Marcos Tidball, Cristina Furlanetto. Meetings of the Brazilian Astronomical Society (SAB), 2021. https://www.eventweb.com.br/specific-files/presentation/sab2021/poster/35.pdf