<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Self-Supervised on Chen Kai Blog</title><link>https://www.chenk.top/zh/tags/self-supervised/</link><description>Recent content in Self-Supervised on Chen Kai Blog</description><generator>Hugo</generator><language>zh-CN</language><lastBuildDate>Wed, 31 Dec 2025 09:00:00 +0000</lastBuildDate><atom:link href="https://www.chenk.top/zh/tags/self-supervised/index.xml" rel="self" type="application/rss+xml"/><item><title>推荐系统（十一）—— 对比学习与自监督学习</title><link>https://www.chenk.top/zh/recommendation-systems/11-%E5%AF%B9%E6%AF%94%E5%AD%A6%E4%B9%A0%E4%B8%8E%E8%87%AA%E7%9B%91%E7%9D%A3%E5%AD%A6%E4%B9%A0/</link><pubDate>Wed, 31 Dec 2025 09:00:00 +0000</pubDate><guid>https://www.chenk.top/zh/recommendation-systems/11-%E5%AF%B9%E6%AF%94%E5%AD%A6%E4%B9%A0%E4%B8%8E%E8%87%AA%E7%9B%91%E7%9D%A3%E5%AD%A6%E4%B9%A0/</guid><description>&lt;p>经典推荐系统只依赖一种信号：用户是否点击、观看或购买？这种信号固然宝贵，却也极其稀疏。大多数用户接触的商品不到总目录的 1%，大多数商品被触达的用户也不到 0.1%，而全新用户或商品则完全没有交互记录。直接用如此稀疏的标签优化模型，几乎注定会在热门头部过拟合，而在长尾部分毫无反应。&lt;/p></description></item></channel></rss>