Results of the summarizing process for the arXiv paper: 1905.13575v1

Solar physics has seen an exponential increase in the volume of data collected over the past decade, and with the introduction of the Daniel K. Inouye Solar Telescope (DKIST), we are entering an age of petabyte solar data. To post-process solar images and create catalogues of data ready for researchers to use, automated feature detection is crucial. A deep learning model proposed by John A. Armstrong and Lyndsay Fletcher can accomplish this task efficiently using a deep convolutional neural network that is adept at feature extraction and image processing. The authors trained their network using data from Hinode/Solar Optical Telescope (SOT) H$\alpha$ images of a small subset of solar features with different geometries such as filaments, prominences, flare ribbons, sunspots, and the quiet Sun (i.e., absence of any other four features). The network achieved near-perfect performance on classifying unseen images from SOT (approximately 99.9%) in just 4.66 seconds. For the first time in a solar context, transfer learning was explored by using pre-trained deep neural networks to help train new models. The authors showed that their network is robust to changes in resolution by degrading images from SOT resolution (approximately 0.33$^{\prime \prime}$ at $\lambda$=6563\AA{}) to Solar Dynamics Observatory/Atmospheric Imaging Assembly (SDO/AIA) resolution (approximately 1.2$^{\prime \prime}$) without affecting its performance. However, it was observed that the network fails to generalize to sunspots from SDO/AIA bands 1600/1700\AA{} due to small-scale brightenings around them and prominences in SDO/AIA 304\AA{} due to coronal emission. Automated feature detection through deep learning models like this one will be invaluable for post-processing massive amounts of solar data collected by DKIST and other solar observatories.

• - Solar physics has seen an exponential increase in data collected over the past decade
• - Introduction of the Daniel K. Inouye Solar Telescope (DKIST) means we are entering an age of petabyte solar data
• - Automated feature detection is crucial for post-processing solar images and creating catalogues of data ready for researchers to use
• - John A. Armstrong and Lyndsay Fletcher proposed a deep learning model using a convolutional neural network to accomplish this task efficiently
• - The authors trained their network using Hinode/Solar Optical Telescope (SOT) H$\alpha$ images of a small subset of solar features with different geometries such as filaments, prominences, flare ribbons, sunspots, and the quiet Sun
• - The network achieved near-perfect performance on classifying unseen images from SOT (approximately 99.9%) in just 4.66 seconds
• - Transfer learning was explored by using pre-trained deep neural networks to help train new models for the first time in a solar context
• - The authors showed that their network is robust to changes in resolution without affecting its performance
• - However, the network fails to generalize to sunspots from SDO/AIA bands 1600/1700\AA{} due to small-scale brightenings around them and prominences in SDO/AIA 304\AA{} due to coronal emission
• - Automated feature detection through deep learning models like this one will be invaluable for post-processing massive amounts of solar data collected by DKIST and other solar observatories

SummaryScientists have collected a lot of information about the sun in the past ten years. They now have a new telescope that will give them even more data. To make sense of all this information, they use computers to help find important features on the sun. Two scientists made a computer program that can quickly and accurately find different parts of the sun. This program can be used to help study the sun and learn more about it. Definitions- Solar physics: The study of the sun and its behavior. - Petabyte: A unit of measurement for digital information storage, equal to one million gigabytes. - Automated feature detection: Using computers to identify important parts or characteristics in images or data automatically. - Convolutional neural network: A type of artificial intelligence algorithm used for image recognition tasks. - Hinode/Solar Optical Telescope (SOT): A telescope used to observe the sun's surface in visible light. - Transfer learning: Using pre-existing knowledge from one task to help solve a new task. - Resolution: The level of detail or clarity in an image or video. - Sunspots: Dark spots on the surface of the sun caused by magnetic activity. - Prominences: Large, bright loops or arcs extending from the sun's surface caused by magnetic fields. - Coronal emission: Radiation emitted by the sun's corona, which is its outermost layer.

Automated Feature Detection with Deep Learning for Solar Data Processing

The field of solar physics has seen an exponential increase in the volume of data collected over the past decade. With the introduction of the Daniel K. Inouye Solar Telescope (DKIST), we are entering an age of petabyte solar data, making automated feature detection crucial for post-processing images and creating catalogues ready for researchers to use. To this end, John A. Armstrong and Lyndsay Fletcher proposed a deep learning model that can efficiently detect features in solar images using a deep convolutional neural network adept at feature extraction and image processing.

Training The Network

The authors trained their network using data from Hinode/Solar Optical Telescope (SOT) H$\alpha$ images containing different geometries such as filaments, prominences, flare ribbons, sunspots, and quiet Sun (i.e., absence of any other four features). The results showed near-perfect performance on classifying unseen images from SOT (approximately 99.9%) in just 4.66 seconds.

Transfer Learning

For the first time in a solar context, transfer learning was explored by using pre-trained deep neural networks to help train new models. This technique proved successful as it allowed them to reduce training time while maintaining accuracy levels comparable to those achieved without transfer learning techniques being used at all.

Robustness To Resolution Changes

The authors also tested their model's robustness to changes in resolution by degrading SOT resolution (approximately 0.33$^{\prime \prime}$ at $\lambda$=6563\AA{}) images down to Solar Dynamics Observatory/Atmospheric Imaging Assembly (SDO/AIA) resolution (approximately 1.2$^{\prime \prime}$). Surprisingly enough, this did not affect its performance significantly; however it was observed that the network fails to generalize well when applied on sunspots from SDO/AIA bands 1600/1700\AA{} due to small-scale brightenings around them or prominences in SDO/AIA 304\AA{} due to coronal emission present there but absent from SOT Hα bandpasses .

Conclusion

Automated feature detection through deep learning models like this one will be invaluable for post-processing massive amounts of solar data collected by DKIST and other solar observatories going forward into the future as more powerful telescopes come online with larger datasets than ever before seen before now possible thanks to advances made in machine learning technology over recent years