diff --git "a/notes/02-\346\267\261\345\272\246\345\255\246\344\271\240\344\273\213\347\273\215.md" "b/notes/02-\346\267\261\345\272\246\345\255\246\344\271\240\344\273\213\347\273\215.md" index 9d51eb8..f99314a 100644 --- "a/notes/02-\346\267\261\345\272\246\345\255\246\344\271\240\344\273\213\347\273\215.md" +++ "b/notes/02-\346\267\261\345\272\246\345\255\246\344\271\240\344\273\213\347\273\215.md" @@ -76,7 +76,7 @@ `x轴`:年份 `y轴`:错误率 `圆点`:表示某年份某研究工作/paper的错误率 [IMAGENET](https://image-net.org/) [数据来源](https://qz.com/1034972/the-data-that-changed-the-direction-of-ai-research-and-possibly-the-world/) -> 在2010年时,错误率比较高,最好的工作错误率也在26%、%27左右; +> 在2010年时,错误率比较高,最好的工作错误率也在26%、27%左右; > > 在2012年,有团队首次使用深度学习将错误率降到25%以下; >