Few-shot learning - Wikipedia
6
公开标注数
6
参与人数
2026-07-22 10:02:21
首次 Whisper
本页的公开 Whisper
划选高亮2026-07-22 16:26:21
原文高亮摘录
“The limiting case of few-shot learning is zero-shot learning that requires no training on examples of the class to be classified”
Whisper 随想笔记
So zero-shot is just few-shot with zero examples, neat way to frame it.
划选高亮2026-07-22 13:23:21
原文高亮摘录
“One-shot learning is the special case of the N-way K-shot framing in which K equals one”
Whisper 随想笔记
Wait, so zero-shot would be K=0? That seems kind of obvious once you say it.
划选高亮2026-07-22 13:14:21
原文高亮摘录
“One-shot learning is the special case of the N-way K-shot framing in which K equals one”
Whisper 随想笔记
So basically one-shot is just a fancy name for K=1, got it.
划选高亮2026-07-22 10:20:21
原文高亮摘录
“Few-shot learning (FSL) is a problem setup in machine learning in which a model learns to perform a task, typically classification, from only a small…”
Whisper 随想笔记
I've tried few-shot learning before and it's way harder than it sounds, but the idea is solid.
划选高亮2026-07-22 10:11:21
原文高亮摘录
“Few-shot learning (FSL) is a problem setup in machine learning in which a model learns to perform a task, typically classification, from only a small…”
Whisper 随想笔记
This is what we need in medical imaging where labeled data is scarce and expensive to get.
划选高亮2026-07-22 10:02:21
原文高亮摘录
“Few-shot learning (FSL) is a problem setup in machine learning in which a model learns to perform a task, typically classification, from only a small…”
Whisper 随想笔记
So it's basically teaching a model with like 5 examples instead of 5,000? Sounds too good to be true.
分享本页 Whisper
短链接
https://domwhisper.com/s/a0b7456778ce嵌入代码
<iframe src="https://domwhisper.com/embed/a0b7456778ce" width="100%" height="480" style="border:0;border-radius:16px" loading="lazy"></iframe>