LOR Deep Probabilist Networks for photoz estimation

Science Photometric Redshifts
#1

Title: LOR Deep Probabilist Networks for photoz estimation
Contributors: Simona Mei, Eric Aubourg, Alexandre Boucaud et Cécile Roucelle
Co-signers: …

  1. Summary Statement

At APC, we intend to use Deep Probabilistic Networks for photoz estimation. The team has a strong expertise in this kind of network. We are starting the research and development of this new tool. Deep Convolutional networks (CNN) have already been very successful for photoz estimation (Shuntov et al. 2020) and probabilistic networks will also be able to give a probability distribution.
A Ph.D. student has started to explore Bayesian CNN and we will continue to develop this kind of method in the next few years.
This is to notify you of what we are doing and we will be able to present more details in the future.

  1. Scientific Utility
    The probabilistic network will extend the work done so far with CNN by giving a probability distribution.

  2. Outputs
    PZ and probability distribution

  3. Performance
    We don’t know yet

  4. Technical Aspects
    N/A