Deep learning approach for identification of HII regions during reionization in 21-cm observations

Abstract

The upcoming Square Kilometre Array (SKA-Low) will map the distribution of neutral hydrogen during reionization, and produce a tremendous amount of 3D tomographic data. These images cubes will be subject to instrumental limitations, such as noise and limited resolution. Here we present SegU-Net, a stable and reliable method for identification of neutral and ionized regions in these images. SegU-Net is a U-Net architecture based convolutional neural network (CNN) for image segmentation. It is capable of segmenting our image data into meaningful features (ionized and neutral regions) with greater accuracy compared to previous methods. We can estimate the true ionization history from our mock observation of SKA with an observation time of 1000 h with more than 87 per cent accuracy. We also show that SegU-Net can be used to recover various topological summary statistics, such as size distributions and Betti numbers, with a relative difference of only a few per cent. These summary statistics characterise the non-Gaussian nature of the reionization process.

arXiv comment15 pages, 13 figures, 3 tables. Moderate and minor revision, consisting of rearranging sections and extend the discussion of the network uncertainty outputs

References in corpus (102)

• Adam: A Method for Stochastic Optimization
• Very Deep Convolutional Networks for Large-Scale Image Recognition
• SciPy 1.0--Fundamental Algorithms for Scientific Computing in Python
• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
• Array Programming with NumPy
• Seven-Year Wilkinson Microwave Anisotropy Probe (WMAP) Observations: Cosmological Interpretation
• scikit-image: Image processing in Python
• Improving neural networks by preventing co-adaptation of feature detectors
• Five-Year Wilkinson Microwave Anisotropy Probe (WMAP) Observations: Cosmological Interpretation
• LOFAR: The LOw-Frequency ARray
• The Effectiveness of Data Augmentation in Image Classification using Deep Learning
• Cosmology at Low Frequencies: The 21 cm Transition and the High-Redshift Universe
• Constraining the Evolution of the Ionizing Background and the Epoch of Reionization with z ~ 6 Quasars II: A Sample of 19 Quasars
• The Murchison Widefield Array: the Square Kilometre Array Precursor at low radio frequencies
• A guide to convolution arithmetic for deep learning
• SparsePak: A Formatted Fiber Field Unit for The WIYN Telescope Bench Spectrograph. I. Design, Construction, and Calibration
• A high-bias, low-variance introduction to Machine Learning for physicists
• Cosmic Reionization and Early Star-Forming Galaxies: A Joint Analysis of New Constraints from Planck and Hubble Space Telescope
• Fully Convolutional Networks for Semantic Segmentation
• 21-cm Tomography of the Intergalactic Medium at High Redshift
• 21cmFAST: A Fast, Semi-Numerical Simulation of the High-Redshift 21-cm Signal
• Statistics of 207 Lya Emitters at a Redshift Near 7: Constraints on Reionization and Galaxy Formation Models
• Hydrogen Epoch of Reionization Array (HERA)
• The Growth of HII Regions During Reionization
• Model-independent evidence in favor of an end to reionization by z~6
• Efficient Simulations of Early Structure Formation and Reionization
• Visualizing and Understanding Convolutional Networks
• What Next-Generation 21 cm Power Spectrum Measurements Can Teach Us About the Epoch of Reionization
• Reionization and the Cosmic Dawn with the Square Kilometre Array
• Quantitative Constraints on the Reionization History from the IGM Damping Wing Signature in Two Quasars at z > 7
• Improved upper limits on the 21-cm signal power spectrum of neutral hydrogen at $\mathbf{z}\approx 9.1$ from LOFAR
• The halo mass function through the cosmic ages
• The Global 21 Centimeter Background from High Redshifts
• 21 cm fluctuations from inhomogeneous X-ray heating before reionization
• Deep multi-redshift limits on Epoch of Reionisation 21cm Power Spectra from Four Seasons of Murchison Widefield Array Observations
• Simulating Cosmic Reionization at Large Scales II: the 21-cm Emission Features and Statistical Signals
• A Method for 21cm Power Spectrum Estimation in the Presence of Foregrounds
• Constraints on reionisation from the z=7.5 QSO ULASJ1342+0928
• The Cosmic Dawn and Epoch of Reionization with the Square Kilometre Array
• High Performance P3M N-body code: CUBEP3M
• 21cmFAST v3: A Python-integrated C code forgenerating 3D realizations of the cosmic 21cm signal
• Evidence of Gunn-Peterson damping wings in high-z quasar spectra: strengthening the case for incomplete reionization
• Quantifying the non-Gaussianity in the EoR 21-cm signal through bispectrum
• The Epoch of Reionization
• Emulating Simulations of Cosmic Dawn for 21cm Power Spectrum Constraints on Cosmology, Reionization, and X-ray Heating
• Constraining the intergalactic medium at $z\approx$ 9.1 using LOFAR Epoch of Reionization observations
• Deep learning from 21-cm tomography of the Cosmic Dawn and Reionization
• The first (nearly) model-independent constraint on the neutral hydrogen fraction at z~5--6
• U-Net: Convolutional Networks for Biomedical Image Segmentation
• Can 21-cm observations discriminate between high-mass and low-mass galaxies as reionization sources?
• Topology and Sizes of HII Regions during Cosmic Reionization
• Automated crater detection on Mars using deep learning
• Analysing the 21cm signal from the Epoch of Reionization with artificial neural networks
• Emulation of reionization simulations for Bayesian inference of astrophysics parameters using neural networks
• Lyman Alpha Emitters at z=7 in the Subaru/XMM-Newton Deep Survey Field: Photometric Candidates and Luminosity Function
• Tight Constraints on the Excess Radio Background at $z=9.1$ from LOFAR
• Measuring the properties of reionised bubbles with resolved Lyman alpha spectra
• Interpreting LOFAR 21-cm signal upper limits at z~9.1 in the context of high-z galaxy and reionisation observations
• Bubble size statistics during reionization from 21-cm tomography
• Emulating the Global 21-cm Signal from Cosmic Dawn and Reionization
• Constraining the astrophysics and cosmology from 21cm tomography using deep learning with the SKA
• Higher Order Contributions to the 21 cm Power Spectrum
• Exploring reionisation and high-z galaxy observables with recent multi-redshift MWA upper limits on the 21-cm signal
• Deep learning dark matter map reconstructions from DES SV weak lensing data
• The Distribution of Bubble Sizes During Reionization
• Imaging the redshifted 21-cm pattern around the first sources during the cosmic dawn using the SKA
• deep21: a Deep Learning Method for 21cm Foreground Removal
• Reionization Through the Lens of Percolation Theory
• Constraining the state of the intergalactic medium during the Epoch of Reionization using MWA 21-cm signal observations
• Evaluating machine learning techniques for predicting power spectra from reionization simulations
• Position-dependent power spectra of the 21-cm signal from the epoch of reionization
• Detecting ionized bubbles in redshifted 21 cm maps
• Recovering the HII region size statistics from 21-cm tomography
• Simulating the Impact of X-ray Heating during the Cosmic Dawn
• Recovering the Wedge Modes Lost to 21-cm Foregrounds
• The impact of spin temperature fluctuations on the 21-cm moments
• Measuring the topology of reionization with Betti numbers
• Evaluating the QSO contribution to the 21-cm signal from the Cosmic Dawn
• Separating the EoR Signal with a Convolutional Denoising Autoencoder: A Deep-learning-based Method
• Neutral island statistics during reionization from 21-cm tomography
• The impact of inhomogeneous subgrid clumping on cosmic reionization
• Nonlinear Bias of Cosmological Halo Formation in the Early Universe
• Removing Astrophysics in 21 cm maps with Neural Networks
• A deep learning model to emulate simulations of cosmic reionization
• Constraining the reionization history using deep learning from 21cm tomography with the Square Kilometre Array
• Prospects of constraining reionization model parameters using Minkowski tensors and Betti numbers
• Tversky loss function for image segmentation using 3D fully convolutional deep networks
• High Precision Analyses of Lyman alpha Damping Wing of Gamma-Ray Bursts in the Reionization Era: On the Controversial Results from GRB 130606A at z = 5.91
• Measuring the History of Cosmic Reionization using the 21-cm PDF from Simulations
• Persistent topology of the reionisation bubble network. I: Formalism & Phenomenology
• Inferring the properties of the sources of reionization using the morphological spectra of the ionized regions
• The impact of inhomogeneous subgrid clumping on cosmic reionization II: modelling stochasticity
• Regularizing Deep Networks with Semantic Data Augmentation
• IslandFAST: A Semi-numerical Tool for Simulating the Late Epoch of Reionization
• Reconstructing Patchy Reionization with Deep Learning
• Bubble mapping with the Square Kilometer Array -- I. Detecting galaxies with Euclid, JWST, WFIRST and ELT within ionized bubbles in the intergalactic medium at z>6
• An Active Galactic Nucleus Recognition Model based on Deep Neural Network
• Using Artificial Neural Networks to Constrain the Halo Baryon Fraction during Reionization
• Predicting 21cm-line map from Lyman $\alpha$ emitter distribution with Generative Adversarial Networks
• Identifying reionization-epoch galaxies with extreme levels of Lyman continuum leakage in James Webb Space Telescope surveys
• Inpainting via Generative Adversarial Networks for CMB data analysis
• Parameters Estimation from the 21 cm signal using Variational Inference

Cited by in corpus (21)

• Machine learning astrophysics from 21 cm lightcones: impact of network architectures and signal contamination
• Exploring the cosmic 21-cm signal from the Epoch of Reionisation using the Wavelet Scattering Transform
• Machine Learning for Observational Cosmology
• Deep learning approach for identification of HII regions during reionization in 21-cm observations -- II. foreground contamination
• BEoRN: A fast and flexible framework to simulate the epoch of reionisation and cosmic dawn
• Supervised machine learning on Galactic filaments Revealing the filamentary structure of the Galactic interstellar medium
• YOLO-CIANNA: Galaxy detection with deep learning in radio data. I. A new YOLO-inspired source detection method applied to the SKAO SDC1
• Detecting the non-Gaussianity of the 21-cm signal during reionisation with the Wavelet Scattering Transform
• Persistent topology of the reionization bubble network. II: Evolution & Classification
• Reconstructing Cosmic Polarization Rotation with ResUNet-CMB
• The problem of dust attenuation in photometric decomposition of edge-on galaxies and possible solutions
• Square Kilometre Array Science Data Challenge 3a: foreground removal for an EoR experiment
• Deep learning approach for identification of HII regions during reionization in 21-cm observations -- III. image recovery
• Topology of Reionisation times: concepts, measurements and comparisons to gaussian random field predictions
• Reionisation time fields reconstruction from 21 cm signal maps
• The 21cm-galaxy cross-correlation: Realistic forecast for 21cm signal detection and reionisation constraints
• Mapping neutral islands during end stages of reionization with photometric intergalactic medium tomography
• CosmoUiT: A Vision Transformer-UNet Hybrid for Fast and Accurate Emulation of 21-cm Maps from the Epoch of Reionization
• A bubble size distribution model for the Epoch of Reionization
• Exploring the Early Universe with Deep Learning
• Nonlinear reconstruction of 21cm global signal from 21cm power spectrum with artificial neural networks