What we do

Astronomy has large amounts of complex data from observations and simulations, which are traditionally used in isolation. This is where we come in.

1. We build a global model of the sky from all available observations.
2. We optimize the full experimental design process from instrumentation to survey strategy and data analysis.

People

Friends

Former members

• Reese Owen (senior thesis)
• Ben Horowitz (postdoc)
• Yunona Iwasaki (junior thesis)
• Satyue Lacayo (junior thesis)
• Andrew Chen (independent work)
• Remy Joseph (postdoc)
• Tianshu Wang (grad student)
• Lucas Makinen (senior thesis)
• Charles Minns (senior thesis)
• Charles Zhao (junior thesis)

Selected Papers

• Score-matching neural networks for improved multi-band source separation
  Sampson, M.; Melchior, P.; Ward, C.; Birmingham, S.
  arXiv:2401.07313
• AESTRA: Deep Learning for Precise Radial Velocity Estimation in the Presence of Stellar Activity
  Liang, Y.; Winn, J.; Melchior, P.
  The Astronomical Journal, 2024, 167, 23
• PopSED: Population-level Inference for Galaxy Properties from Broadband Photometry with Neural Density Estimation
  Li, J.; Melchior, P.; Hahn, C.; Huang, S.
  The Astronomical Journal, 2024, 167, 16
• Autoencoding Galaxy Spectra II: Redshift Invariance and Outlier Detection
  Liang, Y.; Melchior, P.; Lu, S.; Goulding, A.; Ward, C.
  The Astronomical Journal, 2023, 166, 75
• Autoencoding Galaxy Spectra I: Architecture
  Melchior, P.; Liang, Y.; Hahn, C.; Goulding, A.
  The Astronomical Journal, 2023, 166, 74
• Lightweight starshade position sensing with convolutional neural networks and simulation-based inference
  Chen, A.; Harness, A.; Melchior, P.
  Journal of Astronomical Telescopes, Instruments, and Systems, 2023, 9, 025002
• Plausible Adversarial Attacks on Direct Parameter Inference Models in Astrophysics
  Horowitz, B.; Melchior, P.
  arXiv:2211.14788
• Mangrove: Learning Galaxy Properties from Merger Trees
  Jespersen, C.; Cranmer, M.; Melchior, P.; Ho, S.; Somerville, R.; Gabrielpillai, A.
  The Astrophysical Journal, 2022, 941, 7
• Accelerated Bayesian SED Modeling using Amortized Neural Posterior Estimation
  Hahn, C.; Melchior, P.
  The Astrophysical Journal, 2022, 938, 11
• Graph Neural Network-based Resource Allocation Strategies for Multi-Object Spectroscopy
  Wang, T.; Melchior, P.
  Machine Learning: Science and Technology, 2022, 3, 015023
• The challenge of blending in large sky surveys
  Melchior, P.; Joseph, R.; Sanchez, J.; MacCrann, N.; Gruen, D.
  Nature Reviews Physics, 2021, 3, 712
• Joint survey processing: combined resampling and convolution for galaxy modelling and deblending
  Joseph, R.; Melchior, P.; Moolekamp, F.
  arXiv:2107.06984
• Unsupervised Resource Allocation with Graph Neural Networks
  Cranmer, M.; Melchior, P.; Nord, B.
  PMLR, 2021, 148, 272-284
• deep21: a Deep Learning Method for 21cm Foreground Removal
  Makinen, T L; Lancaster, L.; Villaescusa-Navarro, F.; Melchior, P. and 3 co-authors
  JCAP, 2021, 081
• Hybrid Physical-Deep Learning Model for Astronomical Inverse Problems
  Lanusse, F.; Melchior, P.; Moolekamp, F.
  arXiv:1912.03980
• Proximal Adam: Robust Adaptive Update Scheme for Constrained Optimization
  Melchior, P.; Joseph, R.; Moolekamp, F.
  arXiv:1910.10094
• Filling the gaps: Gaussian mixture models from noisy, truncated or incomplete samples
  Melchior, P.; Goulding, A. D.
  A & C, 2018, 25, 183
• SCARLET: Source separation in multi-band images by Constrained Matrix Factorization
  Melchior, P. and 6 co-authors
  A & C, 2018, 24, 129