<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/">
  <channel>
    <title>神经网络 on Co&#39;Log</title>
    <link>https://coldeye2020.github.io/tags/%E7%A5%9E%E7%BB%8F%E7%BD%91%E7%BB%9C/</link>
    <description>Recent content in 神经网络 on Co&#39;Log</description>
    <generator>Hugo -- 0.139.4</generator>
    <language>en</language>
    <lastBuildDate>Wed, 10 Jun 2026 00:00:00 +0000</lastBuildDate>
    <atom:link href="https://coldeye2020.github.io/tags/%E7%A5%9E%E7%BB%8F%E7%BD%91%E7%BB%9C/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>BPQP：加速可微凸优化层的端到端学习框架</title>
      <link>https://coldeye2020.github.io/tech/2026-06-10-bpqp/</link>
      <pubDate>Wed, 10 Jun 2026 00:00:00 +0000</pubDate>
      <guid>https://coldeye2020.github.io/tech/2026-06-10-bpqp/</guid>
      <description>从投资组合优化讲起，系统梳理可微优化层、KKT 条件、隐函数定理、QP/SOCP 求解方法，并解读 NeurIPS 2024 BPQP 如何用一个巧妙的等价转化把反向传播加速一个数量级。</description>
    </item>
  </channel>
</rss>
