{"id":13703894,"url":"https://github.com/wangyuGithub01/Machine_Learning_Resources","last_synced_at":"2025-05-05T09:32:18.817Z","repository":{"id":37451153,"uuid":"173074517","full_name":"wangyuGithub01/Machine_Learning_Resources","owner":"wangyuGithub01","description":":fish::fish::fish: 机器学习面试复习资源","archived":false,"fork":false,"pushed_at":"2023-08-13T21:44:12.000Z","size":17262,"stargazers_count":1122,"open_issues_count":0,"forks_count":174,"subscribers_count":12,"default_branch":"master","last_synced_at":"2024-10-16T00:24:02.067Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":"","language":null,"has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/wangyuGithub01.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null}},"created_at":"2019-02-28T08:41:39.000Z","updated_at":"2024-10-14T02:40:02.000Z","dependencies_parsed_at":"2024-01-14T20:50:13.892Z","dependency_job_id":"ba0204e9-a354-4d0d-91fa-e729b51e31be","html_url":"https://github.com/wangyuGithub01/Machine_Learning_Resources","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/wangyuGithub01%2FMachine_Learning_Resources","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/wangyuGithub01%2FMachine_Learning_Resources/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/wangyuGithub01%2FMachine_Learning_Resources/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/wangyuGithub01%2FMachine_Learning_Resources/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/wangyuGithub01","download_url":"https://codeload.github.com/wangyuGithub01/Machine_Learning_Resources/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":224439571,"owners_count":17311476,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2024-08-02T21:01:01.444Z","updated_at":"2024-11-13T11:30:21.765Z","avatar_url":"https://github.com/wangyuGithub01.png","language":null,"readme":":point_right: 此repo主要是为了整理机器学习面试相关知识点的有用链接\n（注：目前不打算将一些基础算法的内容加入这个repo里，比如LR、SVM算法在《统计学习方法》里已经得到了很好的解释，面试时可能考到的手推公式在书里已经写的很好了，所以推荐直接看书即可。）\n\n### 一、特征工程\n\n#### （一）特征预处理\n- [标准化、归一化、异常特征清洗、不平衡数据](https://www.cnblogs.com/pinard/p/9093890.html)\n- [不平衡数据的处理方法](https://blog.csdn.net/zhang15953709913/article/details/84635540)\n\n#### （二）特征表达\n- [缺失值、特殊特征(eg.时间)、离散特征、连续特征](https://www.cnblogs.com/pinard/p/9061549.html)\n- [连续特征离散化的好处](http://note.youdao.com/noteshare?id=024fa3dbabf4b5a07eb72c8021e60f62)\n- [什么样的模型对缺失值更敏感？](https://blog.csdn.net/zhang15953709913/article/details/88717220)\n\n#### （三）特征选择\n- [过滤法、包装法、嵌入法](https://www.cnblogs.com/pinard/p/9032759.html) \n- [Kaggle中的代码实战](https://www.kaggle.com/willkoehrsen/introduction-to-feature-selection)\n\n### 二、算法基础\n#### （一）评价指标\n- [PR曲线和F1 \u0026 ROC曲线和AUC](http://note.youdao.com/noteshare?id=13d31b4a7dc317b3d4abd18bf42a74df)\n- [AUC \u0026 GAUC](https://zhuanlan.zhihu.com/p/84350940)\n\n#### （二）正则项\n- [正则化与数据先验分布的关系](http://note.youdao.com/noteshare?id=2851b97199bcdc174001d72b1bec0372)\n- [L1在0点处不可导怎么办？](http://www.cnblogs.com/pinard/p/6018889.html)可采用坐标轴下降、最小角回归法\n- [L1为什么比L2的解更稀疏](https://zhuanlan.zhihu.com/p/74874291)\n\n#### （三）损失函数\n- [常见损失函数](https://zhuanlan.zhihu.com/p/58883095)\n- [常见损失函数2](https://zhuanlan.zhihu.com/p/77686118)\n\n#### （四）模型训练\n- [经验误差与泛化误差、偏差与方差、欠拟合与过拟合、交叉验证](http://note.youdao.com/noteshare?id=b629383adb3b09eb31b754c337f690b5)\n- [参数初始化为什么不能全零](https://cloud.tencent.com/developer/article/1535198)\n- [深度学习参数初始化 Lecunn、Xavier、He初始化](https://cloud.tencent.com/developer/article/1542736)\n- [dropout]()\n- [Batch Normalization](https://cloud.tencent.com/developer/article/1551518)\n- [dropout和BN在训练\u0026预测时有什么不同](https://zhuanlan.zhihu.com/p/61725100)\n- [Layer Normalization](https://zhuanlan.zhihu.com/p/113233908)\n- [Transformer为什么用LN不用BN（LN和BN两者分别关注什么）](https://www.zhihu.com/question/395811291/answer/2141681320)\n- [ResNet](https://cloud.tencent.com/developer/article/1591484)\n\n#### （五）优化算法\n- [梯度下降法、牛顿法和拟牛顿法](https://zhuanlan.zhihu.com/p/37524275)\n- [深度学习优化算法SGD、Momentum、Adagrad等](https://zhuanlan.zhihu.com/p/22252270)\n- [最大似然估计 和 最大后验估计](https://zhuanlan.zhihu.com/p/61905474)\n- [最小二乘法 和 最大似然估计的对比联系](https://blog.csdn.net/zhang15953709913/article/details/88716699)\n- [最大似然估计 和 EM](https://blog.csdn.net/zouxy09/article/details/8537620)\n- [浅谈最优化问题的KKT条件](https://zhuanlan.zhihu.com/p/26514613)\n\n#### （六）其他知识点\n- [先验概率 \u0026 后验概率](https://zhuanlan.zhihu.com/p/38567891)\n- [MLE最大似然估计 \u0026 MAP最大后验估计](https://zhuanlan.zhihu.com/p/32480810)\n- [判别模型 vs 生成模型](https://www.zhihu.com/question/20446337)\n- [参数模型 vs 非参数模型](https://zhuanlan.zhihu.com/p/26012348)\n- [参数估计 最大似然估计与贝叶斯估计](https://blog.csdn.net/bitcarmanlee/article/details/52201858)\n- [交叉熵](https://colah.github.io/posts/2015-09-Visual-Information/)\n- [交叉熵 等价 KL散度 等价 MLE最大似然估计](https://zhuanlan.zhihu.com/p/346518942)\n- [向量间距离度量方式](http://note.youdao.com/noteshare?id=ffba716f9f94f1cf3fac48fca300c198)\n- [余弦距离和欧氏距离的转换](https://www.zhihu.com/question/19640394/answer/207795500)\n\n### 三、机器学习算法\n1. 线性回归、逻辑回归、SVM\n  - [LR优缺点](https://github.com/wangyuGithub01/Machine_Learning_Notes/blob/master/pdf/lr_pros_and_cons.md)\n  - [SVM、logistic regression、linear regression对比](https://github.com/wangyuGithub01/Machine_Learning_Notes/blob/master/pdf/compare_svm_lr.md)\n  - [KNN vs K-Means](https://zhuanlan.zhihu.com/p/31580379)\n  - [LR和最大熵模型的关系, LR的并行化](https://blog.csdn.net/dp_BUPT/article/details/50568392)\n  - [为什么LR要用对数似然，而不是平方损失？](https://blog.csdn.net/zhang15953709913/article/details/88717326)\n  - [似然函数](https://zh.wikipedia.org/wiki/%E4%BC%BC%E7%84%B6%E5%87%BD%E6%95%B0)\n2. 树模型\n- [逻辑回归与决策树在分类上的区别](https://blog.csdn.net/zhang15953709913/article/details/84841988)\n- [回归树、提升树、GBDT](https://www.jianshu.com/p/005a4e6ac775)\n- [GBDT、XGBOOST、LightGBM讲解(强烈推荐看一下)](https://github.com/wangyuGithub01/Machine_Learning_Notes/blob/master/pdf/gbdt_wepon.pdf)\n- [XGBOOST具体例子一步步推导，包括缺失值怎么处理（很细值得看）](https://www.jianshu.com/p/ac1c12f3fba1) [(-\u003e 这个链接包含前面文章内容，更全的总结](https://zhuanlan.zhihu.com/p/92837676)\n- [随机森林 GBDT  XGBOOST  LightGBM 比较](http://note.youdao.com/noteshare?id=65790e27fd5737155c31af2c05df8985)\n- [树分裂：信息增益、信息增益率、基尼系数](https://zhuanlan.zhihu.com/p/245617910)\n\n3. 其他\n- [各种机器学习算法的应用场景](https://www.zhihu.com/question/26726794)\n\n### 四、NLP相关\n- word2vec [文章1](https://www.cnblogs.com/pinard/p/7160330.html)[文章2](https://www.cnblogs.com/pinard/p/7243513.html) [文章3](https://www.cnblogs.com/pinard/p/7249903.html)\n- [LSTM](https://zhuanlan.zhihu.com/p/34203833)\n- [LSTM为什么用tanh](https://www.zhihu.com/question/46197687/answer/895834510)\n- [fasttext](https://zhuanlan.zhihu.com/p/32965521)\n- [Transformer、self-attention](https://zhuanlan.zhihu.com/p/54356280)\n- [Transformer图解](https://zhuanlan.zhihu.com/p/338817680)\n- [encode-decode attention和transformer self-attention对比](https://zhuanlan.zhihu.com/p/53682800)\n- [Transformer中的positional encoding](https://www.zhihu.com/question/347678607/answer/864217252)\n- [Bert](https://fancyerii.github.io/2019/03/05/bert-prerequisites/) 零基础入门，prerequisites很全\n- [XLNet](https://zhuanlan.zhihu.com/p/70257427)\n- [nlp中的词向量对比：word2vec/glove/fastText/elmo/GPT/bert](https://zhuanlan.zhihu.com/p/56382372)\n- [NLP/AI面试全记录](https://zhuanlan.zhihu.com/p/57153934)\n\n### 五、推荐系统 \u0026 计算广告 相关\n- [LR \u0026 FTRL](https://zhuanlan.zhihu.com/p/55135954)\n- [FM算法](https://zhuanlan.zhihu.com/p/37963267):讲的蛮细的\n- [FM算法结合推荐系统的讲解](https://zhuanlan.zhihu.com/p/58160982)\n- [DSSM模型](https://zhuanlan.zhihu.com/p/335112207)\n- [DSSM模型的损失函数（顺带讲了point-wise, list-wise, pair-wise损失函数）](https://zhuanlan.zhihu.com/p/322065156)\n- [在线最优化求解 Online Optimization](https://github.com/wzhe06/Ad-papers/blob/master/Optimization%20Method/%E5%9C%A8%E7%BA%BF%E6%9C%80%E4%BC%98%E5%8C%96%E6%B1%82%E8%A7%A3%28Online%20Optimization%29-%E5%86%AF%E6%89%AC.pdf)\n\n### 六、推荐书籍/笔记/代码实现\n- [统计学习方法](https://github.com/wangyuGithub01/E-book/blob/master/%E7%BB%9F%E8%AE%A1%E5%AD%A6%E4%B9%A0%E6%96%B9%E6%B3%95(%E6%9D%8E%E8%88%AA).pdf) (注意这个pdf是第一版，其中的勘误可在[这里](https://github.com/wangyuGithub01/E-book/blob/master/%E7%BB%9F%E8%AE%A1%E5%AD%A6%E4%B9%A0%E6%96%B9%E6%B3%95%EF%BC%88%E5%8B%98%E8%AF%AF%EF%BC%89.pdf)查看)（[代码实现及ppt](https://github.com/fengdu78/lihang-code)）\n- [西瓜书的公式推导细节解析](https://datawhalechina.github.io/pumpkin-book/#/)\n- [deeplearning.ai深度学习课程的中文笔记](https://github.com/fengdu78/deeplearning_ai_books)\n- [机器学习训练秘籍 (Andrew NG)](https://github.com/AcceptedDoge/machine-learning-yearning-cn)\n- [推荐系统实战](https://github.com/wangyuGithub01/E-book)\n\n### 七、推荐专栏\n- [刘建平Pinard](https://www.cnblogs.com/pinard/)：很多高质量文章讲解基础的知识和算法\n- [华校专](http://huaxiaozhuan.com/)：基础算法讲解，多而全（其实还没怎么看\n- [王喆的机器学习专栏](https://zhuanlan.zhihu.com/wangzhenotes)：结合论文+工业界的推荐系统应用，讲的很清晰 \n- [荐道馆](https://www.zhihu.com/column/learningdeep)：讲推荐相关，文章写的比较透\n- [美团技术团队](https://tech.meituan.com/tags/%E7%AE%97%E6%B3%95.html)：美团的技术博客，新技术与实际应用相结合\n- [深度学习前沿笔记](https://zhuanlan.zhihu.com/c_188941548)：NLP相关较多，预训练技术讲解的多\n- [计算广告小觑](https://blog.csdn.net/breada/article/details/50572914)\n- [计算广告论文、学习资料、业界分享](https://github.com/wzhe06/Ad-papers)\n\n### 八、面试问题汇总\n- [牛客网面经总结](https://www.nowcoder.com/discuss/165930)\n\n### 九、其他面试常考\n- [海量数据判重](https://www.nowcoder.com/discuss/153978)\n- [常考智力题/逻辑题](https://github.com/wangyuGithub01/Machine_Learning_Resources/blob/master/pdf/IQ.md)\n- [常考概率题](https://github.com/wangyuGithub01/Machine_Learning_Resources/blob/master/pdf/statistic.md)\n\n### 十、C++相关\n- [STL详解及常见面试题](https://blog.csdn.net/daaikuaichuan/article/details/80717222)\n\n### 工作之后工程实践相关\n\n- [基于PQ量化的近似近邻搜索 (ANN) ](http://xtf615.com/2020/08/01/EBR/)\n- [ANN召回算法之IVFPQ(跟上面的差不多，这篇图第一张画的PQ图更清晰)](https://zhuanlan.zhihu.com/p/378725270)\n","funding_links":[],"categories":["Machine Learning"],"sub_categories":["1. Interview"],"project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FwangyuGithub01%2FMachine_Learning_Resources","html_url":"https://awesome.ecosyste.ms/projects/github.com%2FwangyuGithub01%2FMachine_Learning_Resources","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FwangyuGithub01%2FMachine_Learning_Resources/lists"}