Convolutional neural networks for weak lensing

Convolutional neural networks have revolutionized natural image analysis in the past years. These tools can be also be applied to extract information from astronomical observations or any 2 ( or even more ) dimensional datasets. I will talk about measuring the shapes of galaxies using CNNs with unprecedented speed and precision. I will also present our work on extracting information about cosmology from weak lensing convergence maps, with emphasis on understanding the way the neural network learns to solve the problem.

More Events »