LADDER – Learning Algorithm for Deep Distance Estimation and Reconstruction

Type: Artificial Intelligence applications
Authors: Rahul Shah, Soumadeep Saha, Purba Mukherjee, Utpal Garain, Supratik Pal
Abstract: A deep learning algorithm designed to analyze and reconstruct sequential cosmological data, accounting for covariances between data points, and predicting both mean values and associated uncertainties.

LADDER (Learning Algorithm for Deep Distance Estimation and Reconstruction) is a deep learning framework originally developed to reconstruct the “cosmic distance ladder” by analyzing sequential cosmological data. It utilizes the apparent magnitude data from the Pantheon Type Ia supernovae compilation, fully incorporating covariance information to accurately predict mean values and uncertainties. However, it’s applications extend to other sequential datasets with associated covarince information. Once trained with a particular dataset, e.g. Pantheon, LADDER can have various cosmological applications. It offers model-independent consistency checks for datasets like Baryon Acoustic Oscillations (BAO), and can calibrate high-redshift datasets such as Gamma-Ray Bursts (GRBs) without assuming any underlying cosmological model. Additionally, it serves as a model-independent mock catalog generator for forecast-based cosmological studies. LADDER is aimed at emphasizing the thoughtful integration of machine learning in cosmology.

Link to source: https://github.com/rahulshah1397/LADDER

Related papers: