<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[Aranke的博客]]></title><description><![CDATA[]]></description><link>https://blog.youkuaiyun.com/Aranke</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; Aranke]]></copyright><item><title><![CDATA[【无标题】]]></title><link>https://blog.youkuaiyun.com/Aranke/article/details/166014315</link><guid>https://blog.youkuaiyun.com/Aranke/article/details/166014315</guid><author>Aranke</author><pubDate>Sat, 19 Sep 2026 22:01:50 +0800</pubDate><description><![CDATA[PyTorch核心知识点摘要：高级索引按广播配对，非交叉子矩阵；dim即shape索引，unsqueeze可取ndim，squeeze不可；张量不连续时需.contiguous()或.reshape()；cat拼接、stack堆叠维度不同；梯度默认累加，必须清零；参数更新用.data或no_grad；反向传播中b.grad = dz.sum(dim=0)，W.grad = X.T @ dz，确保形状匹配。]]></description><category></category></item><item><title><![CDATA[笔记-监督学习、特征工程]]></title><link>https://blog.youkuaiyun.com/Aranke/article/details/164376309</link><guid>https://blog.youkuaiyun.com/Aranke/article/details/164376309</guid><author>Aranke</author><pubDate>Sat, 05 Sep 2026 10:28:52 +0800</pubDate><description><![CDATA[机器学习是让计算机从数据中自动学习规律并做出预测或决策的核心技术。本文系统梳理有监督、无监督、自监督、强化学习与半监督学习五大核心范式，分别介绍其核心思想、训练流程与典型应用，并通过Mermaid流程图直观展示整体分类与训练过程，最后简要说明特征工程的基本流程，帮助读者建立从模型训练到特征处理的完整认知。]]></description><category></category></item></channel></rss>