<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[m0_73084903的博客]]></title><description><![CDATA[]]></description><link>https://blog.youkuaiyun.com/m0_73084903</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; m0_73084903]]></copyright><item><title><![CDATA[领域快速入门过程记录之--电力网络]]></title><link>https://blog.youkuaiyun.com/m0_73084903/article/details/150473914</link><guid>https://blog.youkuaiyun.com/m0_73084903/article/details/150473914</guid><author>m0_73084903</author><pubDate>Sun, 17 Aug 2025 23:14:29 +0800</pubDate><description><![CDATA[阅读《Power System Analysis》作为主要参考。]]></description><category></category></item><item><title><![CDATA[GNN--知识图谱（逐步贯通基础到项目实践）]]></title><link>https://blog.youkuaiyun.com/m0_73084903/article/details/149184040</link><guid>https://blog.youkuaiyun.com/m0_73084903/article/details/149184040</guid><author>m0_73084903</author><pubDate>Mon, 07 Jul 2025 23:26:10 +0800</pubDate><description><![CDATA[本文探讨知识图谱构建与跨学科知识关联，重点分析神经网络、图神经网络（GNN）与数学基础（线性代数、图论）的映射关系。主要内容包括：1）线性代数如何支撑神经网络线性层的矩阵运算与并行性；2）图论中的邻接矩阵表示（COO/CSR/CSC）在GNN框架（PYG/DGL）中的应用；3）两大框架的硬件加速策略对比：PYG基于COO格式的稀疏矩阵优化，DGL采用CSR/CSC格式的块级并行与算子融合。通过知识图谱可视化与实例分析，揭示了前置基础学科对深度学习技术的支撑作用，以及框架设计如何反哺对数学原理的深层理解。]]></description><category></category></item><item><title><![CDATA[线性代数--AI数学基础复习]]></title><link>https://blog.youkuaiyun.com/m0_73084903/article/details/149156925</link><guid>https://blog.youkuaiyun.com/m0_73084903/article/details/149156925</guid><author>m0_73084903</author><pubDate>Sun, 06 Jul 2025 19:10:39 +0800</pubDate><description><![CDATA[参考：麻省理工公开课 线性代数 MIT Linear Algebra。]]></description><category></category></item><item><title><![CDATA[【无标题】]]></title><link>https://blog.youkuaiyun.com/m0_73084903/article/details/149037686</link><guid>https://blog.youkuaiyun.com/m0_73084903/article/details/149037686</guid><author>m0_73084903</author><pubDate>Tue, 01 Jul 2025 08:18:49 +0800</pubDate><description><![CDATA[本文讨论了GNN中结构信息输入方式的改进方法，重点分析了GNN-PE论文的创新点。传统GNN通过多层消息传递学习结构信息存在局限，论文提出三种解决方案：叠加层数、高阶节点元组聚合和位置编码(PE)。作者指出当前图神经网络缺乏规范化的PE方法，现有结合Transformer和GNN的方法中PE存在不可学习或符号模糊问题。论文创新性地提出结构位置和结构表示，让PE拥有独立更新网络并与隐状态合并更新。除拉普拉斯PE外，还引入随机游走PE，其有效性取决于不同图结构所需的最优游走步数k。]]></description><category></category></item><item><title><![CDATA[计算机系统结构]]></title><link>https://blog.youkuaiyun.com/m0_73084903/article/details/147069944</link><guid>https://blog.youkuaiyun.com/m0_73084903/article/details/147069944</guid><author>m0_73084903</author><pubDate>Tue, 08 Apr 2025 17:05:30 +0800</pubDate><description><![CDATA[经典定义：传统机器程序员所看到的计算机属性，即概念性结构与功能特性。按照计算机系统结构的多级层次结构，不同程序员看到的计算机具有不同的属性。底层机器的属性对于高层机器的程序员来说往往是透明的。广义定义：指令系统结构、组成、硬件（实现）实质就是确定计算机系统中的软硬件界面，界面之上时软件实现的功能，界面之下时硬件和固件实现的功能。]]></description><category></category></item><item><title><![CDATA[编译原理课设]]></title><link>https://blog.youkuaiyun.com/m0_73084903/article/details/145944175</link><guid>https://blog.youkuaiyun.com/m0_73084903/article/details/145944175</guid><author>m0_73084903</author><pubDate>Sat, 01 Mar 2025 10:48:24 +0800</pubDate><description><![CDATA[目标：passcal-S语言编译程序的设计与实现。]]></description><category></category></item><item><title><![CDATA[Kaggler日志--Day9]]></title><link>https://blog.youkuaiyun.com/m0_73084903/article/details/144589458</link><guid>https://blog.youkuaiyun.com/m0_73084903/article/details/144589458</guid><author>m0_73084903</author><pubDate>Thu, 19 Dec 2024 19:36:25 +0800</pubDate><description><![CDATA[昨日复盘：补充并解决Day7统计的部分问题今日进度：继续完成Day8统计问题的解答。]]></description><category></category></item><item><title><![CDATA[Kaggler日志--Day8]]></title><link>https://blog.youkuaiyun.com/m0_73084903/article/details/144548853</link><guid>https://blog.youkuaiyun.com/m0_73084903/article/details/144548853</guid><author>m0_73084903</author><pubDate>Wed, 18 Dec 2024 19:37:23 +0800</pubDate><description><![CDATA[对Day6代码的详细分析，解决其中问题，理解完整工作流中需要明白的步骤的关键概念]]></description><category></category></item><item><title><![CDATA[Kaggler日志--Day7]]></title><link>https://blog.youkuaiyun.com/m0_73084903/article/details/144543114</link><guid>https://blog.youkuaiyun.com/m0_73084903/article/details/144543114</guid><author>m0_73084903</author><pubDate>Tue, 17 Dec 2024 23:47:47 +0800</pubDate><description><![CDATA[昨日复盘：尝试自己爬取了两个学校的就业信息数据，比较简单但是顺通了爬虫流程看别人的代码：AQX的。今日进度：分析理解昨天代码的过程，统计问题。]]></description><category></category></item><item><title><![CDATA[Kaggler日志--Day6]]></title><link>https://blog.youkuaiyun.com/m0_73084903/article/details/144517675</link><guid>https://blog.youkuaiyun.com/m0_73084903/article/details/144517675</guid><author>m0_73084903</author><pubDate>Mon, 16 Dec 2024 20:35:37 +0800</pubDate><description><![CDATA[昨日复盘：今日进度：尝试自己爬取了两个学校的就业信息数据，比较简单但是顺通了爬虫流程看别人的代码：AQX的。比较详细，内容也比较多，慢慢看。]]></description><category></category></item><item><title><![CDATA[Kaggler日志--Day5]]></title><link>https://blog.youkuaiyun.com/m0_73084903/article/details/144490923</link><guid>https://blog.youkuaiyun.com/m0_73084903/article/details/144490923</guid><author>m0_73084903</author><pubDate>Sun, 15 Dec 2024 22:44:43 +0800</pubDate><description><![CDATA[昨日复盘之类型变量读两篇讲解如何提问的文章，在提问区里发起一次提问实战：自己从头到尾首先并成功提交之管道（pipeline之前一直错译为工作流）今日进度之交叉验证之XGBoost之数据泄露利用以上所学刷一遍分数。]]></description><category></category></item><item><title><![CDATA[Kaggler日志-Day4]]></title><link>https://blog.youkuaiyun.com/m0_73084903/article/details/144470733</link><guid>https://blog.youkuaiyun.com/m0_73084903/article/details/144470733</guid><author>m0_73084903</author><pubDate>Sat, 14 Dec 2024 17:47:35 +0800</pubDate><description><![CDATA[Pandas课程完成2/7之类型变量读两篇讲解如何提问的文章，在提问区里发起一次提问实战：自己从头到尾首先并成功提交。]]></description><category></category></item><item><title><![CDATA[Kaggler日志-Day3]]></title><link>https://blog.youkuaiyun.com/m0_73084903/article/details/144454855</link><guid>https://blog.youkuaiyun.com/m0_73084903/article/details/144454855</guid><author>m0_73084903</author><pubDate>Fri, 13 Dec 2024 20:32:12 +0800</pubDate><description><![CDATA[完成Panda学习。
开始进行深度学习中级课程，学习了缺失值的处理]]></description><category></category></item><item><title><![CDATA[Kaggler日志-Day2]]></title><link>https://blog.youkuaiyun.com/m0_73084903/article/details/144415771</link><guid>https://blog.youkuaiyun.com/m0_73084903/article/details/144415771</guid><author>m0_73084903</author><pubDate>Thu, 12 Dec 2024 22:08:42 +0800</pubDate><description><![CDATA[Kaggle入门每日记录]]></description><category></category></item><item><title><![CDATA[数据结构记录]]></title><link>https://blog.youkuaiyun.com/m0_73084903/article/details/137608696</link><guid>https://blog.youkuaiyun.com/m0_73084903/article/details/137608696</guid><author>m0_73084903</author><pubDate>Wed, 10 Apr 2024 20:18:45 +0800</pubDate><description><![CDATA[前端搭建：Vue + Element。]]></description><category></category></item><item><title><![CDATA[图神经网络实践00--参考PYG文档]]></title><link>https://blog.youkuaiyun.com/m0_73084903/article/details/136891176</link><guid>https://blog.youkuaiyun.com/m0_73084903/article/details/136891176</guid><author>m0_73084903</author><pubDate>Wed, 20 Mar 2024 23:05:37 +0800</pubDate><description><![CDATA[通过文档学习PYG。]]></description><category></category></item><item><title><![CDATA[图神经网络论文解读理论00--distill论文入手]]></title><link>https://blog.youkuaiyun.com/m0_73084903/article/details/136843617</link><guid>https://blog.youkuaiyun.com/m0_73084903/article/details/136843617</guid><author>m0_73084903</author><pubDate>Tue, 19 Mar 2024 22:41:49 +0800</pubDate><description><![CDATA[distill是网页形式的期刊，其中有许多非常棒的可视化讲解，力求把晦涩难懂的研究工作讲得通俗易懂。非常适合初学者学习。其中有两篇发布在的有关图神经网络的博客，文章脉络清晰。以下将参考这两篇论文展开。]]></description><category></category></item><item><title><![CDATA[用numpy搭建自己的神经网络]]></title><link>https://blog.youkuaiyun.com/m0_73084903/article/details/136439834</link><guid>https://blog.youkuaiyun.com/m0_73084903/article/details/136439834</guid><author>m0_73084903</author><pubDate>Sun, 03 Mar 2024 23:08:53 +0800</pubDate><description><![CDATA[构建模型的基本思想：构建深度学习的过程：产生idea，将idea转化成code，最后进行experiment，之后根据结果修改idea，继续idea–>code–>experiment的循环，直到最终训练到表现不错的深度学习网络模型。]]></description><category></category></item><item><title><![CDATA[计算机网络]]></title><link>https://blog.youkuaiyun.com/m0_73084903/article/details/136374381</link><guid>https://blog.youkuaiyun.com/m0_73084903/article/details/136374381</guid><author>m0_73084903</author><pubDate>Thu, 29 Feb 2024 22:25:51 +0800</pubDate><description><![CDATA[计算机网络：由单一技术相连的自主计算机集合。（独立设备之间的信息交换）分布式系统：一系列独立计算机作为一个独立统一系统呈现。（一个常见的分布式系统是World Wide Web，in which everything looks like a document (web page)）]]></description><category></category></item><item><title><![CDATA[计算机组成原理]]></title><link>https://blog.youkuaiyun.com/m0_73084903/article/details/136347086</link><guid>https://blog.youkuaiyun.com/m0_73084903/article/details/136347086</guid><author>m0_73084903</author><pubDate>Wed, 28 Feb 2024 16:46:09 +0800</pubDate><description><![CDATA[计算机系统是一个由硬件、软件组成的复杂的自动化设备。]]></description><category></category></item></channel></rss>