. NET Platform Machine learning

. NET Platform Machine learning Resources Summary, do you have what you want?

Contact machine learning for more than 1 years, because it will only use C # heap code, so only focus. NET platform of resources, while accumulating, while collecting, while learning, so in this site 101th blog arrival, share to everyone. Some of the used, there will be a little more detailed instructions, other useless, but also I am concerned about, perhaps later will be used. Machine learning is not the same as big data or data mining, there are some differences, some things can be used to deal with big data problems or data mining problems, they also have some figured out, so these components can not only be used for machine learning, but also for data mining related.

According to the function of the resources are divided into 3 parts, open source and non-comprehensive categories, as well as other website blog information. are able to be in . NET platform is used. Thank you for your support, these components I will certainly study its use in the future, then I would have to share again. If you are interested, you can follow this blog.

This text address: http://www.cnblogs.com/asxinyu/p/4422050.html

1. Open source Comprehensive class 1.1aforge.net

Aforge.net is designed for developers and researchers based on the C # framework, including computer vision and artificial intelligence, image processing, neural networks, genetic algorithms, machine learning, fuzzy systems, robot control and other fields. This framework consists of a series of class libraries. Mainly includes:

aforge.imaging--some daily image processing and filters
aforge.vision--Computer Vision Application Class Library
aforge.neuro--Neural Network Computing Library aforge.genetic-Evolutionary algorithm programming library
aforge.machinelearning--Machine Learning Class Library
aforge.robotics--provides a library of machine learning tool classes
aforge.video--series of video processing class libraries
aforge.fuzzy--Fuzzy Inference System class Library
aforge.controls--image, three-dimensional, chart display controls

From: http://baike.haosou.com/doc/1786119-1888850.html

official website : http://www.aforgenet.com/

Personally, I think this is. NET Platform for machine learning and data mining the longest, best, most comprehensive open source. One of the. NET components. Blog Park has a lot of friends have written a special use of the article. I am also just concerned, not yet used, because the direction and processing of the problem is not the same, for the time being no practical application. Source code, cases, etc. are very comprehensive.

1.2 Accord.net Framework

The accord.net framework is encapsulated and further developed on a aforge.net basis. The function is also very powerful, because Aforge.net is more focused on some underlying and breadth, while the accord.net framework focuses more on the machine learning profession, on the basis of which provides more statistical analysis and processing functions, including image processing and computer vision algorithms, so the focus is different, But they are very useful.

official website : http://accord-framework.net/

1.3 math.net

Whether it is machine learning or data mining, and mathematics inseparable from the relationship, since it is in. NET platform, then you might be using this component later. Math.net is one of the most comprehensive mathematical computing components under the. NET Platform, and the basic functionality is perfect. My blog has a detailed study of this component:http://www.cnblogs.com/asxinyu/p/4329737.html . Of course, more features have to be used by everyone to explore, after all, provide the source code. Math.net was originally designed to open up a stable and continuously maintained, advanced Basic Math toolkit to meet the daily needs of. NET developers. Currently, the component is divided into the following sub-projects, which also support mono, and the supported platforms are very broad. Math.net Numerics is the core function of numerical computation. It is mainly to provide daily scientific engineering calculation related algorithms, including some special functions, linear algebra, probability theory, stochastic functions, calculus, interpolation, optimization and other related computational functions. Detailed introduction and use can refer to the menu "math.net" of the site to view the directory.

official website : http://www.mathdotnet.com/

1.4 Infer.net

Well, the ones above are very powerful, strong on the one hand is said to include a wide range, on the one hand is code, annotations, resources, case is also very perfect. If the above is cannon, then I think this infer.net is fighter, fragmented contact and research, and translation of its document code has been 5 months, the longer, the more it feels the power of its firepower. My blog has published 2 translated documents: http://www.cnblogs.com/asxinyu/p/4329742.html, please pay attention.

Infer.net is a machine inference component developed by Shine based on the. NET platform that uses Microsoft Research License Agreement authorization, non-commercial use Only.Infer.NET is a framework for running a Bayesian inference machine (Bayesian inference) in a probabilistic graph model (graphical models). If you do not know the meaning of the probability map model or Bayesian inference, you can refer to the Resources and References page pages. Infer.net provides advanced messaging algorithms and statistical procedures for the reasoning required by various applications. Infer.net is more concerned with the application of probability map programming or Bayesian theory. This random factor and many problems in the uncertain world can be applied, so his strength is focused on the one hand, and the other is the modeling language provided. Unlike other components, other components are algorithm-level, while infer.net is a modeling level, with a variety of common and common inference algorithms. The model can be created by simple code, which, according to Microsoft, is the MSL modeling language, and this is where the component gives me an awe, and it is estimated that only Microsoft researchers would think of doing it once and for all.

official website : http://research.microsoft.com/en-us/um/cambridge/projects/infernet/default.aspx

1.5 NUML

Another small, multi-machine learning algorithm class library that supports supervised and unsupervised learning. Many common machine learning algorithms are supported, and documentation resources are good. including Cluster,kmeans,pca,decisiontree,knn,naivebayes,neuralnetwork and other learning algorithms, the content is also very rich, powerful, but also includes some numerical calculation of the implementation. This component personally thinks that it is not as complex as the above, the structure is small and reasonable, and the code is elegant. Take a look at the code below and quickly build a decision tree learner to make predictions:

1 var generator = new Decisiontreegenerator (); 2 generator. Descriptor = descriptor.create<tennis> (); 3 Generator. Sethint (FALSE); 4  5 tennis[] tennis = Tennisdata.getdata (); 6  7 var learned = Learner.learn (tennis, 0.80, generator); 8  9 IModel model = learned. model;10 double accuracy = learned.  ACCURACY;11 Tennis t = new Tennis13 {     Outlook = outlook.sunny,15     temperature = temperature.high,16     Windy = False17};18 Tennis predictedval = model. Predict (t);

　NUML's introductory case and documentation are comprehensive, and if you are familiar with the algorithm, familiarity with C #, then getting started should not be a problem. And you can build and solve more complex problems through the components themselves.

official website : http://numl.net/

1.6 Alglib

Alglib is a cross-platform numerical analysis and data processing function library that includes both open source and commercial versions. It supports a variety of programming languages, such as C++,C#,PASCAL,VBA, which can be run on multiple operating system platforms, such as Windows,linux and Solaris. Alglib has the following features:

(1) linear algebra (including matrix analysis);
(2) equation solving (linear and nonlinear);
(3) interpolation;
(4) Optimization;
(5) Fast Fourier transform;
(6) numerical integration;
(7) linear and nonlinear least squares fitting;
(8) Solving the ordinary differential equation;
(9) Special functions;
(10) Statistics (descriptive statistics, hypothesis testing);
(11) Data analysis (classification, regression, neural network);

official website : http://www.alglib.net/

2. Open source. NET platform non-comprehensive class 2.1 AdaBoost algorithm

1.https://github.com/bgorven/classifier

2.https://github.com/elmerning/adaboost

2.2 Apriori algorithm

1.https://github.com/omar-salem/apriori-algorithm

2.https://github.com/simonesalvo/apriori

2.3 PageRank algorithm

Https://github.com/archgold/pagerank

2.4 Nativebayes (naive Bayesian) algorithm

1.https://github.com/rekin/naive-bayes-classifier
2.https://github.com/ardaxi/bayes.net
3.https://github.com/amrishdeep/dragon
4.https://github.com/joelmartinez/nbayes

2.5 Kmeans algorithm

Http://visualstudiomagazine.com/articles/2013/12/01/k-means-data-clustering-using-c.aspx

3. Other resources and technology blogs

"Resources" 108 Big Data document PDF open Download-After finishing package download, although the relevant information of big data, mainly PPT, but there is a little relationship with machine learning, need to see;

Vernacular Bayesian theory and its application in the prediction of football match results and C # Implementation of the "attached data" there are Bayesian related papers, the article itself on the naïve Bayesian principle is also introduced very clearly;

Research on ten classical algorithms in Data mining field

10 Classic Algorithms for machine learning

A popular introduction to support vector machines (understanding the three-tier realm of SVM)

The Bayesian classification algorithm, EM and HMM are discussed from the decision tree learning.

Natural Language Processing blog, contains a lot of content, maybe a bit strong theoretical http://www.52nlp.cn/

Northwest University of Technology Bo guide Nie Fei Ping Blog: http://www.escience.cn/people/fpnie/index.html

A machine learning Data mining blog, there are a lot of resources link: http://www.zhizhihu.com/

Http://mlg.eng.cam.ac.uk/mlss09/schedule.htm

A centralized platform for machine learning resources Http://www.kernel-machines.org/software

Algorithm grocer--naive Bayesian classification of classification algorithm (Naive Bayesian classification)

Introduction to Maximum entropy model

Probabilistic graph model

Blog Park Bobby0322 Blog: http://www.cnblogs.com/Bobby0322/p/4052495.html in the Business intelligence and data Mining applications series articles:

1. A preliminary study on data mining of "Bi's little Thing"
2. The main method of data mining in the bi thing
3. "Bi thing" analysis of 13 kinds of commonly used data mining technology
4. The relationship between data mining and related fields in Bi thing
5. "Bi Thing" Microsoft Association algorithm
6. "Bi thing" Microsoft Clustering algorithm
7. "Bi Thing" Microsoft cluster analysis algorithm--three-person identity division
8. "Bi thing" Microsoft Decision Tree algorithm
9. "Bi thing" Microsoft decision Tree algorithm--Find out the characteristics of the Three Kingdoms, dedicated to the broad three enthusiasts
10. "Bi thing" Microsoft linear regression algorithm
11. "Bi thing" Microsoft logistic regression algorithm
12. "Bi thing" Microsoft logistic regression algorithm--predicting the rise and fall of stocks
13. "Bi thing" Microsoft Naive Bayes algorithm
14. "Bi thing" Microsoft Naive Bayes algorithm--three-person identity division
15. "Bi thing" Microsoft neural network algorithm
16. "Bi thing" Microsoft sequential analysis and clustering algorithm
17. "Bi Thing" Microsoft time Series algorithm
18. "Bi Thing" Microsoft time Series algorithm--Verifying the magical Fibonacci sequence
19. "Bi thing" data mining algorithms--verification of accuracy

4. My 100 Blog Paths

It has been more than 5 years since the blog Park was registered on August 1, 2009. This is the 100th official essay article of the blog. In the 2015 New Year's Day, watching their blog for a long time did not update, only more than 40 articles, and then listed a writing plan, initially is at least finished high-quality articles about 50. And now only 4 months, I did not think I almost completed the year-round goal. Of course more than 50 articles published, I think high quality and meaningful may only be 40, but it does not matter, at least a lot of time. These articles are a summary of their own experience and knowledge, but also an improvement. In these 100 blog milestones, I briefly reviewed these 100 articles.

1th article on the first day of browsing to 1000:

Xcode uses records-automatically inserts test data into the database using Xcode (2012-04-25 09:11)

1th article on the first day of browsing to 3000:

Have your own code generator-newlife.xcode Template Authoring Tutorial (2012-05-11 08:35)

The 1th article on the blog Park headlines:

Challenge ORM Performance--newlife.xcode 5 million SQLite database paging (2012-08-22-12:22)

The 1th article recommends more than 60 articles:

"Original" open source Word read-write component docx Introduction and Getting Started (2013-02-22 10:35) 24

The 1th article recommends more than 80 articles:

"5.1 Gifts" domestic first matlab and C #. NET Mixed Programming video tutorial "Free" (2014-04-29 08:02)

1th article of total Views: over 16,000

"Spit Groove" VS2012 installation project can only be used InstallShield Limited Edition (2013-09-07-11:20)

Of all the 100 essays, 13 are directories and link summaries, not a written essay, and 9 articles are just beginning to blog Park when, still learning, technical content is not high. But I did not delete, after all, is a history. Plus there are 2 articles about bitcoin and source code, not exactly what I wrote, most of it is the content of the big stone, there are 2 resources and Baidu Spit Groove is very casual writing, there is no intention to publish at home, just to do a record. So the actual comparison has a technical point of the article or the number of experience is 73. In these 73 articles:

there are at least 15 articles in the blog Park headline that I think is a good one . (includes " most recommended " and " Max comments " and " Edit recommendations ")

1. Vernacular Bayesian theory and its application in the prediction of soccer match results and C # implementation "attached data"

2. Have you ever used this wonderful C # note? How to look at it (2015-04-17-10:04)

3. "Original" C # play high-frequency digital color fast 3 of a little experience (2015-04-11-09:03)

4. "Stepping on the Pit" once ASP. Small Web site Deployment pit and settlement experience (2015-04-01 06:10)

5. "Share" blog Landscaping (4) Add a smart article to the blog recommendation plugin (2015-03-24-07:55)

6. Common functions of "original" Newlife.xcode use (a) query and data initialization (2015-01-26-08:52)

7. "Original" open source math.net basic Math Library use C # to implement other random number generators (2015-03-18-08:32)

8. "Anti-MLM" MLM Story summary-How to protect themselves and their families as much as possible (2015-03-09 07:37)

9. "Anti-pyramid marketing" a short-term misunderstanding of the marketing and escape of the true story and the Thinking of technology (2015-03-03-06:10)

10.App troubled times, 3721 how far away from us (2015-02-10 09:24)

11. "Original" open source Word read-write component docx Introduction and Getting Started (2013-02-22 10:35)

12. "Original" C # Open Source Lightweight Object Database Ndatabase Introduction (2013-02-20-09:35)

13. "Original". NET open source compression components introduction and Getting Started (2013-03-05-07:59)

14. "5.1 Gifts" domestic first matlab and C #. NET Mixed Programming video tutorial "Free" (2014-04-29 08:02)

There is also an article by the blog Garden as an editorial recommendation article:

15. "Original" matlab.net mixed programming techniques direct call to Matlab built-in functions

In General, the article is very efficient and recognized by everyone, although the technical content is not particularly advanced, but perhaps the basis of the technology more can resonate it. What I want to say is that each article is carefully edited and written out, the results are very ideal, got a lot of people's support and understanding, so there is so efficient visits and recommendations and comments.

If you feel that reading this article is helpful to you, please click " recommend " button, your "recommendation" will be my biggest writing motivation! Welcome to reprint, but without the author's consent, reprint article must be in the article page obvious location to the author and the original link , otherwise reserve the right to pursue legal responsibility.

. NET Platform Machine learning