Ziye Yu, Yuqi Cai, Weitao Wang, Yanru An, Lu Li, Yueyang Xia, Yunpeng Zhang
{"title":"PRIME-DP:地震数据处理预训练综合模型","authors":"Ziye Yu, Yuqi Cai, Weitao Wang, Yanru An, Lu Li, Yueyang Xia, Yunpeng Zhang","doi":"arxiv-2408.01919","DOIUrl":null,"url":null,"abstract":"We introduce a new seismic wave representation model called PRIME-DP, which\nstands for Pre-trained Integrated Model for Earthquake Data Processing. Unlike\nmost of the models, which are designed to specifically a singular problem,\nPRIME-DP is used for multi-task single station seismic waveform processing.\nPRIME-DP can be used to Pg/Sg/Pn/Sn phase picking, P polarization\nclassification. And can be fine-tunned to wide range of tasks, such as event\nclassification, without architecture modifications. PRIME-DP can achieve over\n85% recall on Pg and Sg phases, when picking continuous waveform and achieves\nover 80% accuracy in P polarization classification. By fine-tunning\nclassification decoder with NeiMeng dataset, PRIME-DP can achieve 95.1%\naccuracy on event.","PeriodicalId":501270,"journal":{"name":"arXiv - PHYS - Geophysics","volume":"48 1","pages":""},"PeriodicalIF":0.0000,"publicationDate":"2024-08-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"PRIME-DP: Pre-trained Integrated Model for Earthquake Data Processing\",\"authors\":\"Ziye Yu, Yuqi Cai, Weitao Wang, Yanru An, Lu Li, Yueyang Xia, Yunpeng Zhang\",\"doi\":\"arxiv-2408.01919\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"We introduce a new seismic wave representation model called PRIME-DP, which\\nstands for Pre-trained Integrated Model for Earthquake Data Processing. Unlike\\nmost of the models, which are designed to specifically a singular problem,\\nPRIME-DP is used for multi-task single station seismic waveform processing.\\nPRIME-DP can be used to Pg/Sg/Pn/Sn phase picking, P polarization\\nclassification. And can be fine-tunned to wide range of tasks, such as event\\nclassification, without architecture modifications. PRIME-DP can achieve over\\n85% recall on Pg and Sg phases, when picking continuous waveform and achieves\\nover 80% accuracy in P polarization classification. By fine-tunning\\nclassification decoder with NeiMeng dataset, PRIME-DP can achieve 95.1%\\naccuracy on event.\",\"PeriodicalId\":501270,\"journal\":{\"name\":\"arXiv - PHYS - Geophysics\",\"volume\":\"48 1\",\"pages\":\"\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-08-04\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"arXiv - PHYS - Geophysics\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/arxiv-2408.01919\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"arXiv - PHYS - Geophysics","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/arxiv-2408.01919","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
摘要
我们引入了一种新的地震波表示模型,称为 PRIME-DP,即地震数据处理预训练综合模型(Pre-trained Integrated Model for Earthquake Data Processing)。PRIME-DP 可用于 Pg/Sg/Pn/Sn 相位选取、P 极化分类。PRIME-DP 可用于 Pg/Sg/Pn/Sn 相位拾取、P 偏振分类,并可在不修改架构的情况下对事件分类等多种任务进行微调。在挑选连续波形时,PRIME-DP 对 Pg 和 Sg 相位的召回率超过 85%,对 P 极化分类的准确率超过 80%。通过使用内蒙古数据集对分类解码器进行微调,PRIME-DP 在事件分类上的准确率达到 95.1%。
PRIME-DP: Pre-trained Integrated Model for Earthquake Data Processing
We introduce a new seismic wave representation model called PRIME-DP, which
stands for Pre-trained Integrated Model for Earthquake Data Processing. Unlike
most of the models, which are designed to specifically a singular problem,
PRIME-DP is used for multi-task single station seismic waveform processing.
PRIME-DP can be used to Pg/Sg/Pn/Sn phase picking, P polarization
classification. And can be fine-tunned to wide range of tasks, such as event
classification, without architecture modifications. PRIME-DP can achieve over
85% recall on Pg and Sg phases, when picking continuous waveform and achieves
over 80% accuracy in P polarization classification. By fine-tunning
classification decoder with NeiMeng dataset, PRIME-DP can achieve 95.1%
accuracy on event.