DeepSphere: Efficient spherical Convolutional Neural Network with HEALPix sampling for cosmological applications

Abstract

Convolutional Neural Networks (CNNs) are a cornerstone of the Deep Learning toolbox and have led to many breakthroughs in Artificial Intelligence. These networks have mostly been developed for regular Euclidean domains such as those supporting images, audio, or video. Because of their success, CNN-based methods are becoming increasingly popular in Cosmology. Cosmological data often comes as spherical maps, which make the use of the traditional CNNs more complicated. The commonly used pixelization scheme for spherical maps is the Hierarchical Equal Area isoLatitude Pixelisation (HEALPix). We present a spherical CNN for analysis of full and partial HEALPix maps, which we call DeepSphere. The spherical CNN is constructed by representing the sphere as a graph. Graphs are versatile data structures that can act as a discrete representation of a continuous manifold. Using the graph-based representation, we define many of the standard CNN operations, such as convolution and pooling. With filters restricted to being radial, our convolutions are equivariant to rotation on the sphere, and DeepSphere can be made invariant or equivariant to rotation. This way, DeepSphere is a special case of a graph CNN, tailored to the HEALPix sampling of the sphere. This approach is computationally more efficient than using spherical harmonics to perform convolutions. We demonstrate the method on a classification problem of weak lensing mass maps from two cosmological models and compare the performance of the CNN with that of two baseline classifiers. The results show that the performance of DeepSphere is always superior or equal to both of these baselines. For high noise levels and for data covering only a smaller fraction of the sphere, DeepSphere achieves typically 10% better classification accuracy than those baselines. Finally, we show how learned filters can be visualized to introspect the neural network.

arXiv commentarXiv admin note: text overlap with arXiv:astro-ph/0409513 by other authors

References in corpus (35)

• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
• Planck 2015 results. XIII. Cosmological parameters
• Seven-Year Wilkinson Microwave Anisotropy Probe (WMAP) Observations: Cosmological Interpretation
• HEALPix -- a Framework for High Resolution Discretization, and Fast Analysis of Data Distributed on the Sphere
• The Emerging Field of Signal Processing on Graphs: Extending High-Dimensional Data Analysis to Networks and Other Irregular Domains
• Geometric deep learning: going beyond Euclidean data
• The clustering of galaxies in the completed SDSS-III Baryon Oscillation Spectroscopic Survey: cosmological analysis of the DR12 galaxy sample
• Weak Gravitational Lensing
• Planck 2013 results. I. Overview of products and scientific results
• Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering
• HI4PI: A full-sky HI survey based on EBHIS and GASS
• KiDS-450: Cosmological parameter constraints from tomographic weak gravitational lensing
• Convolutional Networks on Graphs for Learning Molecular Fingerprints
• The Fourteenth Data Release of the Sloan Digital Sky Survey: First Spectroscopic Data from the extended Baryon Oscillation Spectroscopic Survey and from the second phase of the Apache Point Observatory Galactic Evolution Experiment
• Dark Energy Survey Year 1 Results: Cosmological Constraints from Cosmic Shear
• A Convolutional Neural Network for Modelling Sentences
• Gravitational Lensing
• Stationary signal processing on graphs
• Learning shape correspondence with anisotropic convolutional neural networks
• Learning Spherical Convolution for Fast Features from 360° Imagery
• Non-Gaussian information from weak lensing data via deep learning
• Deep learning from 21-cm tomography of the Cosmic Dawn and Reionization
• Dark Energy Survey Year 1 Results: Curved-Sky Weak Lensing Mass Map
• Libsharp - spherical harmonic transforms revisited
• Cosmological constraints from noisy convergence maps through deep learning
• Approximate fast graph Fourier transforms via multi-layer sparse approximations
• Machine learning cosmological structure formation
• Detecting Cosmic Strings in the CMB with the Canny Algorithm
• Convolutional Neural Networks on the HEALPix sphere: a pixel-based algorithm and its application to CMB data analysis
• Cosmological constraints from the convergence 1-point probability distribution
• Fast Generation of Covariance Matrices for Weak Lensing
• A Bayesian Framework for Cosmic String Searches in CMB Maps
• Estimating Cosmological Parameters from the Dark Matter Distribution
• Harmonic Inpainting of the Cosmic Microwave Background Sky: Formulation and Error Estimate
• Reionization Models Classifier using 21cm Map Deep Learning

Cited by in corpus (70)

• WeatherBench: A benchmark dataset for data-driven weather forecasting
• Learning to Predict the Cosmological Structure Formation
• Gauge Equivariant Convolutional Networks and the Icosahedral CNN
• Cosmological constraints with deep learning from KiDS-450 weak lensing maps
• The DAWES review 10: The impact of deep learning for the analysis of galaxy surveys
• Weak lensing cosmology with convolutional neural networks on noisy data
• Applications and Techniques for Fast Machine Learning in Science
• DeepCMB: Lensing Reconstruction of the Cosmic Microwave Background with Deep Neural Networks
• Parameters Estimation for the Cosmic Microwave Background with Bayesian Neural Networks
• A Full $w$CDM Analysis of KiDS-1000 Weak Lensing Maps using Deep Learning
• Large Scale Graph Learning from Smooth Signals
• The GCE in a New Light: Disentangling the $\gamma$-ray Sky with Bayesian Graph Convolutional Neural Networks
• Full-sky Cosmic Microwave Background Foreground Cleaning Using Machine Learning
• Vertex and Energy Reconstruction in JUNO with Machine Learning Methods
• Convolutional Neural Networks on the HEALPix sphere: a pixel-based algorithm and its application to CMB data analysis
• Gauge Equivariant Mesh CNNs: Anisotropic convolutions on geometric graphs
• A neural simulation-based inference approach for characterizing the Galactic Center $\gamma$-ray excess
• A Hybrid Deep Learning Approach to Cosmological Constraints From Galaxy Redshift Surveys
• Probabilistic Mass Mapping with Neural Score Estimation
• Direction of Arrival Estimation of Sound Sources Using Icosahedral CNNs
• A General Theory of Equivariant CNNs on Homogeneous Spaces
• Natural Graph Networks
• Dim but not entirely dark: Extracting the Galactic Center Excess' source-count distribution with neural nets
• OSLO: On-the-Sphere Learning for Omnidirectional images and its application to 360-degree image compression
• Recovering the CMB Signal with Machine Learning
• The Power of Halometry
• Guided Co-Modulated GAN for 360° Field of View Extrapolation
• Cosmic Velocity Field Reconstruction Using AI
• Neural Networks Optimized by Genetic Algorithms in Cosmology
• Decimated Framelet System on Graphs and Fast G-Framelet Transforms
• Spherical Multi-Modal Place Recognition for Heterogeneous Sensor Systems
• DeepSphere: a graph-based spherical CNN
• Identification of Patterns in Cosmic-Ray Arrival Directions using Dynamic Graph Convolutional Neural Networks
• Ground state search by local and sequential updates of neural network quantum states
• A Generative Model of Galactic Dust Emission Using Variational Inference
• Cosmological Parameter Estimation and Inference using Deep Summaries
• DeepSphere: towards an equivariant graph-based spherical CNN
• Cosmological Particle Production and Pairwise Hotspots on the CMB
• Deep learning the astrometric signature of dark matter substructure
• Identifying nearby sources of ultra-high-energy cosmic rays with deep learning
• Coordinate Independent Convolutional Networks -- Isometry and Gauge Equivariant Convolutions on Riemannian Manifolds
• Spin-Weighted Spherical CNNs
• Wavelet-Based Segmentation on the Sphere
• Quantum algorithms for group convolution, cross-correlation, and equivariant transformations
• A tomographic spherical mass map emulator of the KiDS-1000 survey using conditional generative adversarial networks
• Scattering Networks on the Sphere for Scalable and Rotationally Equivariant Spherical CNNs
• Planet cartography with neural learned regularization
• Deep learning approach for identification of HII regions during reionization in 21-cm observations -- III. image recovery
• Classifying CMB time-ordered data through deep neural networks
• Riemannian Data preprocessing in Machine Learning to focus on QCD color structure
• Cosmic Kite: Auto-encoding the Cosmic Microwave Background
• Deep Needlet: A CNN based full sky component separation method in Needlet space
• Accelerating MCMC algorithms through Bayesian Deep Networks
• Self-supervised component separation for the extragalactic submillimeter sky
• CMB delensing with deep learning
• Spherical Transformer: Adapting Spherical Signal to CNNs
• Spherical Image Generation from a Single Normal Field of View Image by Considering Scene Symmetry
• Efficient Generalized Spherical CNNs
• Capacity of Group-invariant Linear Readouts from Equivariant Representations: How Many Objects can be Linearly Classified Under All Possible Views?
• Conditional Generative Modeling via Learning the Latent Space
• Spherical Convolutional Neural Networks: Stability to Perturbations in SO(3)
• Spherical Sliced-Wasserstein
• On the Energy Distribution of the Galactic Center Excess' Sources
• Equivariant Networks for Pixelized Spheres
• Advancing weak lensing mass mapping with a mask-aware HEALPix transformer
• D${}_4$CNN$\times$AnaCal: Physics-Informed Machine Learning for Accurate and Precise Weak Lensing Shear Estimation
• Score-based generative emulation of impact-relevant Earth system model outputs
• Towards detecting Primordial non-Gaussianity in the CMB using Spherical Convolutional Neural Networks
• A study of an energy-dependent anisotropy of cosmic rays beyond the GZK cut-off with deep neural networks
• ChebLieNet: Invariant Spectral Graph NNs Turned Equivariant by Riemannian Geometry on Lie Groups