{"title":"A versatile framework for analyzing galaxy image data by incorporating Human-in-the-loop in a large vision model*","authors":"Ming-Xiang Fu, 溟翔 傅, Yu Song, 宇 宋, Jia-Meng Lv, 佳蒙 吕, Liang Cao, 亮 曹, Peng Jia, 鹏 贾, Nan Li, 楠 李, Xiang-Ru Li, 乡儒 李, Ji-Feng Liu, 继峰 刘, A-Li Luo, 阿理 罗, Bo Qiu, 波 邱, Shi-Yin Shen, 世银 沈, Liang-Ping Tu, 良平 屠, Li-Li Wang, 丽丽 王, Shou-Lin Wei, 守林 卫, Hai-Feng Yang, 海峰 杨, Zhen-Ping Yi, 振萍 衣, Zhi-Qiang Zou and 志强 邹","doi":"10.1088/1674-1137/ad50ab","DOIUrl":null,"url":null,"abstract":"The exponential growth of astronomical datasets provides an unprecedented opportunity for humans to gain insight into the Universe. However, effectively analyzing this vast amount of data poses a significant challenge. In response, astronomers are turning to deep learning techniques, but these methods are limited by their specific training sets, leading to considerable duplicate workloads. To overcome this issue, we built a framework for the general analysis of galaxy images based on a large vision model (LVM) plus downstream tasks (DST), including galaxy morphological classification, image restoration, object detection, parameter extraction, and more. Considering the low signal-to-noise ratios of galaxy images and the imbalanced distribution of galaxy categories, we designed our LVM to incorporate a Human-in-the-loop (HITL) module, which leverages human knowledge to enhance the reliability and interpretability of processing galaxy images interactively. The proposed framework exhibits notable few-shot learning capabilities and versatile adaptability for all the abovementioned tasks on galaxy images in the DESI Legacy Imaging Surveys. In particular, for the object detection task, which was trained using 1000 data points, our DST in the LVM achieved an accuracy of 96.7%, while ResNet50 plus Mask R-CNN reached an accuracy of 93.1%. For morphological classification, to obtain an area under the curve (AUC) of ~0.9, LVM plus DST and HITL only requested 1/50 of the training sets that ResNet18 requested. In addition, multimodal data can be integrated, which creates possibilities for conducting joint analyses with datasets spanning diverse domains in the era of multi-messenger astronomy.","PeriodicalId":10250,"journal":{"name":"中国物理C","volume":"46 1","pages":""},"PeriodicalIF":3.6000,"publicationDate":"2024-08-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"中国物理C","FirstCategoryId":"101","ListUrlMain":"https://doi.org/10.1088/1674-1137/ad50ab","RegionNum":2,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"PHYSICS, NUCLEAR","Score":null,"Total":0}
引用次数: 0
Abstract
The exponential growth of astronomical datasets provides an unprecedented opportunity for humans to gain insight into the Universe. However, effectively analyzing this vast amount of data poses a significant challenge. In response, astronomers are turning to deep learning techniques, but these methods are limited by their specific training sets, leading to considerable duplicate workloads. To overcome this issue, we built a framework for the general analysis of galaxy images based on a large vision model (LVM) plus downstream tasks (DST), including galaxy morphological classification, image restoration, object detection, parameter extraction, and more. Considering the low signal-to-noise ratios of galaxy images and the imbalanced distribution of galaxy categories, we designed our LVM to incorporate a Human-in-the-loop (HITL) module, which leverages human knowledge to enhance the reliability and interpretability of processing galaxy images interactively. The proposed framework exhibits notable few-shot learning capabilities and versatile adaptability for all the abovementioned tasks on galaxy images in the DESI Legacy Imaging Surveys. In particular, for the object detection task, which was trained using 1000 data points, our DST in the LVM achieved an accuracy of 96.7%, while ResNet50 plus Mask R-CNN reached an accuracy of 93.1%. For morphological classification, to obtain an area under the curve (AUC) of ~0.9, LVM plus DST and HITL only requested 1/50 of the training sets that ResNet18 requested. In addition, multimodal data can be integrated, which creates possibilities for conducting joint analyses with datasets spanning diverse domains in the era of multi-messenger astronomy.
期刊介绍:
Chinese Physics C covers the latest developments and achievements in the theory, experiment and applications of:
Particle physics;
Nuclear physics;
Particle and nuclear astrophysics;
Cosmology;
Accelerator physics.
The journal publishes original research papers, letters and reviews. The Letters section covers short reports on the latest important scientific results, published as quickly as possible. Such breakthrough research articles are a high priority for publication.
The Editorial Board is composed of about fifty distinguished physicists, who are responsible for the review of submitted papers and who ensure the scientific quality of the journal.
The journal has been awarded the Chinese Academy of Sciences ‘Excellent Journal’ award multiple times, and is recognized as one of China''s top one hundred key scientific periodicals by the General Administration of News and Publications.