memect / hao

好东西传送门
1.4k stars 459 forks source link

@刘赛的中国梦:如何在机器学习的语境下能达到有坚实的数学基础? #81

Closed haoawesome closed 10 years ago

haoawesome commented 10 years ago

同求: Coder_Chenzhi: 我想补习机器学习涉及到的统计学知识,但是相关内容太多,不知道该如何下手,可以推荐一些资源吗?谢谢! http://www.weibo.com/message/history?uid=1693756354&name=Coder_Chenzhi

haoawesome commented 10 years ago

http://stats.stackexchange.com/questions/40808/is-a-strong-background-in-maths-a-total-requisite-for-ml Is a strong background in maths a total requisite for ML?

haoawesome commented 10 years ago

http://www.csml.ucl.ac.uk/courses/msc_ml/?q=node/111 General math pre-requisites

• Calculus • Linear Algebra • Statistics/Probability • Stochastic processes and dynamical systems

haoawesome commented 10 years ago

Machine Learning: How can you learn Mathematics for machine learning? http://www.quora.com/Machine-Learning/How-can-you-learn-Mathematics-for-machine-learning

Matrix Cookbook http://www.quora.com/Machine-Learning/How-can-you-learn-Mathematics-for-machine-learning

Linear Algebra class at MIT http://ocw.mit.edu/courses/mathematics/18-06-linear-algebra-spring-2010/video-lectures/

haoawesome commented 10 years ago

CMU course called Computer Science Theory for the Information Age http://www.cs.cmu.edu/~venkatg/teaching/CStheory-infoage/ Notes 1: Geometry of high-dimensional space Notes 2: Singular Value Decomposition Notes 3: Algorithms for Massive Data Sets Notes 4: Random graphs (see Chapter 3 of draft of whole book) Notes 5: Markov chains Notes 6: Learning theory (see Chapter 6 of draft of whole book) Notes 7: Clustering

The book http://www.cs.cmu.edu/~venkatg/teaching/CStheory-infoage/hopcroft-kannan-feb2012.pdf

haoawesome commented 10 years ago

Christopher Bishop - Pattern Recognition & Machine Learning http://research.microsoft.com/en-us/um/people/cmbishop/prml/

haoawesome commented 10 years ago

Stanford University Machine Learning Andrew Ng https://www.coursera.org/course/ml

haoawesome commented 10 years ago

Stanford University Probabilistic Graphical Models Daphne Koller https://www.coursera.org/course/pgm

haoawesome commented 10 years ago

http://courses.washington.edu/css490/2012.Winter/lecture_slides/02_math_essentials.pdf Machine Learning Math Essentials

LiuSai-Dream commented 10 years ago

自问自答,根据这些天的收获,我推荐一本书,《数据挖掘中的新方法-支持向量机(邓乃扬)》这本书有点早,其里面以svm为重点,当然也对其他方面进行了拓展,例如期望风险涉及到损失函数和概率分布。 本书的不足之处在于没有提及到smo优化算法,而是详细介绍了传统的优化算法,例如牛顿法,变尺度法,gbfs... 读了这本书在读《统计学习方法》就会比较容易了

haoawesome commented 10 years ago

数据挖掘中的新方法-支持向量机(邓乃扬) http://product.dangdang.com/8874033.html

http://pan.baidu.com/wap/shareview?&shareid=1438288027&uk=2466605404&dir=%2Fartificial%20intelligence%2Fmachine%20learning%2Fstatistical%20learning%20theory%2FSVM%2FSVM%E7%9A%84%E6%96%B0%E8%BF%9B%E5%B1%95%EF%BC%88%E5%8F%82%E8%80%83%E6%96%87%E7%8C%AE%EF%BC%89&page=1&num=20&fsid=2503190986&third=0

李航 《统计学习方法》 http://book.douban.com/subject/10590856/

haoawesome commented 10 years ago

@Coder_Chenzhi 您好,我想补习机器学习涉及到的统计学知识,但是相关内容太多,不知道该如何下手,可以推荐一些资源吗?谢谢!

你可以关注这个问答追踪: https://github.com/memect/hao/issues/81 我们会尽快把答案整理好, 也欢迎你描述自己的背景,以便我们有针对性地推荐. 感谢你的支持

coder-chenzhi commented 10 years ago

研一新生,现在跟着师兄做实验,他让我实现用Dirichlet Process实现聚类数目自动确定,但是本科只学了简单的《概率论和数理统计》,对DP完全没概念,所以从DP这一点一直往上溯源,Gibbs Sampling、Markov Chain Monte Carlo、Probabilistic Graphical Models、Bayersian Inference等等,然后就混乱了,不知该从哪个开始看。。。谢谢解答!

haoawesome commented 10 years ago

@coder-chenzhi 大致分几步

  1. 了解一些相关技术基础知识, 看短教程/讲义/教科书, 先大致明白,知道将来到哪里查资料就行 https://www.cs.cmu.edu/~kbe/dp_tutorial.pdf 看看这个CMU 的 Dirichlet Process 短教程
  2. 搞清楚目标问题, 最好能找到一个具体的领域问题,有数据,可以试验(这样容易发文章) 如果实在没有想法,可以找几篇最近几年的顶级会议的相关文章, 看看他们有没有缺陷,技巧,发展目标.
haoawesome commented 10 years ago

初学者问题:如何在机器学习的语境下能达到有坚实的数学基础? 讨论见 http://t.cn/RPlPk6o @刘赛的中国梦 提到《数据挖掘中的新方法-支持向量机》(邓乃扬) 读了这本书再读 @李航博士 《统计学习方法》就会比较容易了。此外CMU等名校的机器学习课程都列有前提要求,可资参考 http://www.weibo.com/5220650532/Btv9Nuvqd

haoawesome commented 10 years ago

上微博的猫V 个人体会:机器学习的语境下,很难提高数学基础(本来数学好的另说)。数学和机器学习的世界观是不一样的。还是老老实实看数学分析、线性代数、统计学以及最优化,比较靠谱。 http://www.weibo.com/1679022231/Btvk93TSA?ref=atme

haoawesome commented 10 years ago

@民工_李江 回复@好东西传送门:我现在的理解,很难在数学领域划个区域,说只要搞定这部分数学就能通吃ML了。所以大部分人都是边学边补。如果硬要推荐的话,统计很多人推荐“All of Statistics”,线代都推荐Strang的“Introduction to Linear Algebra”。

haoawesome commented 10 years ago

All of Statistics http://www.amazon.com/All-Statistics-Statistical-Inference-Springer/dp/0387402721 pdf http://read.pudn.com/downloads158/ebook/702714/Larry%20Wasserman_ALL%20OF%20Statistics.pdf

haoawesome commented 10 years ago

Introduction to Linear Algebra http://www.amazon.com/Introduction-Linear-Algebra-Fourth-Edition/dp/0980232716 http://astro.temple.edu/~tue79412/books/linearAlgebra.pdf