A Distributed Learning Architecture for Scientific Imaging Problems

 

Authors: A. Panousopoulou, S. Farrens, K. Fotiadou, A. Woiselle, G. Tsagkatakis, J-L. Starck,  P. Tsakalides
Journal: arXiv
Year: 2018
Download: ADS | arXiv


Abstract

Current trends in scientific imaging are challenged by the emerging need of integrating sophisticated machine learning with Big Data analytics platforms. This work proposes an in-memory distributed learning architecture for enabling sophisticated learning and optimization techniques on scientific imaging problems, which are characterized by the combination of variant information from different origins. We apply the resulting, Spark-compliant, architecture on two emerging use cases from the scientific imaging domain, namely: (a) the space variant deconvolution of galaxy imaging surveys (astrophysics), (b) the super-resolution based on coupled dictionary training (remote sensing). We conduct evaluation studies considering relevant datasets, and the results report at least 60\% improvement in time response against the conventional computing solutions. Ultimately, the offered discussion provides useful practical insights on the impact of key Spark tuning parameters on the speedup achieved, and the memory/disk footprint.

Weak Lensing 2D and 3D Density Fluctuation Map Reconstruction

Weak lensing 2D & 3D density fluctuation map reconstruction


The 3D tomographic weak lensing is one of the most important tools for modern cosmology:  Underlying the link between weak lensing and the compressed sensing theory, we have proposed a  new approach to reconstruct the dark matter distribution in two and three dimensions, using photometric redshift information. We have shown  that we can estimate with a very good accuracy the mass and redshift of dark matter haloes, which is crucial for unveiling the nature of the Dark Universe (Leonard et al. 2014). We have shown that it outperforms significantly all existing methods. In particular, we have seen using simulations that we can reconstruct two clusters on the same light of sight, which was impossible with previous methods.  The method has be chosen by the DES consortium to general its weak lensing mass map (Jeffrey et al, 2018).

Reference 1: A. Leonard, F. Lanusse and J.-L. Starck, "GLIMPSE: Accurate 3D weak lensing reconstructions using sparsity", MNRAS, 440, 2, 2014.

Reference 2: F. Lanusse, J.-L. Starck, A. Leonard, S. Pires, "High Resolution Weak Lensing Mass-Mapping Combining Shear and Flexion", Astronomy and Astrophysics, 591, id.A2, 19 pp, 2016.

Reference 3: Niall Jeffrey et al., MNRAS, 479, 2018, arXiv:1801.08945.

Press release: CEA press release

 

Cosmology and Fundamental Physics with the Euclid Satellite

Cosmology and Fundamental Physics with the Euclid Satellite


Understanding the source of cosmic acceleration in the universe is one of the major challenges that will be addressed by future surveys like the Euclid space mission. Acceleration may be caused by a cosmological constant or by a dynamical fluid (dark energy) or rather be a sign that the laws of gravity themselves are different at very large scales. Euclid data interpretation will aim at discriminating among these scenarios. CosmoStat is active in the Theory Working Group, and V.Pettorino led the update of the Review Cosmology and Fundamental Physics with the Euclid Satellite, published on Living Reviews in Relativity in 2018 https://link.springer.com/article/10.1007%2Fs41114-017-0010-3. The figure below (originally from Hu and Sawicki (2007a), replotted as Fig.19 of the Review, shows constraints expected for Euclid on the growth factor, for different cosmological scenarios.

Reference : Euclid Theory Working Group, Cosmology and Fundamental Physics with the Euclid Satellite,

DOI: https://doi.org/10.1007/s41114-017-0010-3 published on 12 April 2018.

DEDALE: Mathematical Tools to Help Navigate the Big Data Maze

Managing the huge volumes and varying streams of Big Data digital information presents formidable analytical challenges to anyone wanting to make sense of it. Consider the mapping of space, where scientists collect, process and transmit giga-scale data sets to generate accurate visual representations of millions of galaxies. Or consider the vast information being generated by genomics and bioinformatics as genomes are mapped and new drugs discovered. And soon the Internet of Things will bring millions of interconnected information-sensing and transmitting devices.