2000 character limit reached
Score-matching neural networks for improved multi-band source separation
Published 14 Jan 2024 in astro-ph.IM | (2401.07313v2)
Abstract: We present the implementation of a score-matching neural network that represents a data-driven prior for non-parametric galaxy morphologies. The gradients of this prior can be incorporated in the optimization of galaxy models to aid with tasks like deconvolution, inpainting or source separation. We demonstrate this approach with modification of the multi-band modeling framework Scarlet that is currently employed as deblending method in the pipelines of the HyperSuprimeCam survey and the Rubin Observatory. The addition of the prior avoids the requirement of non-differentiable constraints, which can lead to convergence failures we discovered in Scarlet. We present the architecture and training details of our score-matching neural network and show with simulated Rubin-like observations that using a data-driven prior outperforms the baseline \scarlet method in accuracy of total flux and morphology estimates, while maintaining excellent performance for colors. We also demonstrate significant improvements in the robustness to inaccurate initializations. The trained score models used for this analysis are publicly available at https://github.com/SampsonML/galaxygrad.
- Posterior samples of source galaxies in strong gravitational lenses with score-based priors, in: Machine Learning and the Physical Sciences Workshop, p. E1. doi:10.48550/arXiv.2211.03812, arXiv:2211.03812.
- Reverse-time diffusion equation models. Stochastic Processes and their Applications 12, 313–326.
- Deblending galaxies with variational autoencoders: A joint multiband, multi-instrument approach. Monthly Notices of the Royal Astronomical Society 500, 531–547. doi:10.1093/mnras/staa3062, arXiv:2005.12039.
- The Zwicky Transient Facility: System Overview, Performance, and First Results. Publications of the Astronomical Society of the Pacific 131, 018002. doi:10.1088/1538-3873/aaecbe.
- SExtractor: Software for source extraction. Astronomy and Astrophysics Supplement Series 117, 393–404. doi:10.1051/aas:1996164.
- The Hyper Suprime-Cam software pipeline. Publications of the Astronomical Society of Japan 70, S5. doi:10.1093/pasj/psx080, arXiv:1705.06766.
- Photometry of high-redshift blended galaxies using deep learning. Monthly notices of the Royal Astronomical Society 491, 2481–2495. URL: https://academic-oup-com.ezproxy.princeton.edu/mnras/article-pdf/491/2/2481/31196214/stz3056.pdf, doi:10.1093/mnras/stz3056.
- Deblending and classifying astronomical sources with Mask R-CNN deep learning. Monthly Notices of the Royal Astronomical Society 490, 3952–3965. doi:10.1093/mnras/stz2845, arXiv:1908.02748.
- A proximal decomposition method for solving convex variational inverse problems. Inverse problems URL: http://iopscience.iop.org/article/10.1088/0266-5611/24/6/065014/meta.
- Proximal splitting methods in signal processing URL: http://arxiv.org/abs/0912.3522, arXiv:0912.3522.
- The Zwicky Transient Facility: Observing System. Publications of the Astronomical Society of the Pacific 132, 038001. doi:10.1088/1538-3873/ab4ca2.
- The zwicky transient facility: Science objectives. Publications of the Astronomical Society of the Pacific 131, 1–23. doi:10.1088/1538-3873/ab006c.
- Deblending Galaxies with Generative Adversarial Networks. The Astrophysical Journal 941, 141. doi:10.3847/1538-4357/aca1b8, arXiv:2211.04488.
- Denoising Diffusion Probabilistic Models. arXiv e-prints , arXiv:2006.11239arXiv:2006.11239.
- LSST: From Science Drivers to Reference Design and Anticipated Data Products. The Astrophysical Journal 873, 111. doi:10.3847/1538-4357/ab042c, arXiv:0805.2366.
- Multi-band morpho-Spectral Component Analysis Deblending Tool (MuSCADeT): Deblending colourful objects. Astronomy and Astrophysics 589, A2. doi:10.1051/0004-6361/201527923, arXiv:1603.00473.
- Hybrid Physical-Deep learning model for astronomical inverse problems URL: http://arxiv.org/abs/1912.03980, arXiv:1912.03980.
- Sdss image processing i : The deblender. URL: https://api.semanticscholar.org/CorpusID:85560178.
- Dark Energy Survey Y3 results: blending shear and redshift biases in image simulations. Monthly Notices of the Royal Astronomical Society 509, 3371–3394. doi:10.1093/mnras/stab287010.48550/arXiv.2012.08567, arXiv:2012.08567.
- Weak Lensing for Precision Cosmology. Annual Review of Astronomy and Astrophysics 56, 393–433. doi:10.1146/annurev-astro-081817-05192810.48550/arXiv.1710.03235, arXiv:1710.03235.
- Proximal adam: Robust adaptive update scheme for constrained optimization. arXiv e-prints , arXiv:1910.10094URL: https://ui.adsabs.harvard.edu/abs/2019arXiv191010094M.
- The challenge of blending in large sky surveys. Nature Reviews Physics 3, 712–718. doi:10.1038/s42254-021-00353-y.
- SCARLET: Source separation in multi-band images by Constrained Matrix Factorization. Astronomy and Computing 24, 129. doi:10.1016/j.ascom.2018.07.001, arXiv:1802.10157.
- Blendingtoolkit. URL: https://lsstdesc.org/BlendingToolKit/index.html.
- Block-simultaneous direction method of multipliers: a proximal primal-dual splitting algorithm for nonconvex problems with multiple constraints. Optimization and Engineering. International Multidisciplinary Journal to Promote Optimization Theory & Applications in Engineering Sciences URL: https://doi.org/10.1007/s11081-018-9380-y, doi:10.1007/s11081-018-9380-y.
- Improved Denoising Diffusion Probabilistic Models. arXiv e-prints , arXiv:2102.09672arXiv:2102.09672.
- Galaxy blending effects in deep imaging cosmic shear probes of cosmology. Monthly Notices of the Royal Astronomical Society 514, 5905–5926. doi:10.1093/mnras/stac130310.48550/arXiv.2112.07659, arXiv:2112.07659.
- Proximal algorithms. Foundations and Trends® in Optimization 1, 127–239. URL: http://dx.doi.org/10.1561/2400000003, doi:10.1561/2400000003.
- Deblending galaxy superpositions with branched generative adversarial networks. Monthly Notices of the Royal Astronomical Society 485, 2617–2627. doi:10.1093/mnras/stz575, arXiv:1810.10098.
- GALSIM: The modular galaxy image simulation toolkit. Astronomy and Computing 10, 121–150. doi:10.1016/j.ascom.2015.02.002, arXiv:1407.7676.
- Spotting Hallucinations in Inverse Problems with Data-Driven Priors. arXiv e-prints , arXiv:2306.13272doi:10.48550/arXiv.2306.13272, arXiv:2306.13272.
- Super-Convergence: Very fast training of neural networks using large learning rates URL: http://arxiv.org/abs/1708.07120, arXiv:1708.07120.
- Realistic galaxy image simulation via score-based generative models. Monthly Notices of the Royal Astronomical Society 511, 1808–1818. doi:10.1093/mnras/stac130, arXiv:2111.01713.
- Generative modeling by estimating gradients of the data distribution. Advances in Neural Information Processing Systems 32.
- Improved Techniques for Training Score-Based Generative Models. arXiv e-prints , arXiv:2006.09011arXiv:2006.09011.
- Score-Based Generative Modeling through Stochastic Differential Equations. arXiv e-prints , arXiv:2011.13456arXiv:2011.13456.
- Galaxy deblending using residual dense neural networks. Physical Review D 106, 063023. doi:10.1103/PhysRevD.106.063023, arXiv:2109.09550.
- Panic at the ISCO: time-varying double-peaked broad lines from evolving accretion disks are common amongst optically variable AGN. arXiv e-prints , arXiv:2309.02516doi:10.48550/arXiv.2309.02516, arXiv:2309.02516.
Paper Prompts
Sign up for free to create and run prompts on this paper.
