H11-M02.zip
From the dataset abstract
GradientHide 透過新增公共資料更新步驟並使用 CLIP 進行標籤對齊來防止聯邦學習中的梯度反轉，從而有效地保護隱私，同時保持跨基準資料集的模型準確性。
Source: H11-M02_隱私保護之聯邦式學習模型
Additional Information
| Field | Value |
|---|---|
| Data last updated | October 17, 2025 |
| Metadata last updated | September 30, 2025 |
| Created | September 30, 2025 |
| Format | application/zip |
| License | Other (Non-Commercial) |
| created | 2 months ago |
| format | ZIP |
| id | 1e3a094f-32f1-443e-b625-4f7b4ce9092d |
| last modified | 1 month ago |
| md5 | 1429fe7ec7eaee21d63e16c688d7f4c0 |
| mimetype | application/zip |
| on same domain | True |
| package id | ab7ef4cd-36b2-42c6-b73c-9590c2b37ce0 |
| proxy url | https://scidm.nchc.org.tw/en/dataset/ab7ef4cd-36b2-42c6-b73c-9590c2b37ce0/resource/1e3a094f-32f1-443e-b625-4f7b4ce9092d/nchcproxy/H11-M02.zip |
| revision id | 1f4c2520-7f06-4b75-af68-787207f8fefa |
| sha256 | ab33406a06f6bededb43924f6c2ef18b0bee643a53a9ae10cd7b2ecaa6645715 |
| size | 239.3 KiB |
| state | active |
| url type | upload |
推薦資料集：