Machine Learning Materials

1. Information on the bottom of the press box

There are also some commonly collected information on the bottom of the press box to share with you:

1.1 IPOL - C implementation of classic computer vision algorithms

http://www.ipol.im/?utm_source=doi

1.2 https://www.codecademy.com/ ——Online programming self-taught

My python is self-taught on this website. Basically, the basic data structure of python, list, dist, etc. is introduced. It only takes about a week or even less time to basically master a new language.

1.3 Draw a block diagram online - the best option without visio

https://www.processon.com/

If there is no visio this is the best choice!

1.4 Some great bloggers

Liu Weipeng
http://mindhacks.cn/
http://mindhacks.cn/2011/11/04/how-to-interview-a-person-for-two-years/
http://mindhacks.cn/2012/08 /27/modern-cpp-practices/

Liao Xuefeng's python tutorial
https://www.liaoxuefeng.com/wiki/0014316089557264a6b348958f449949df42a6d3a2e542c000

1.5 Options for writing technical blogs

I have been working in csdn for more than 8 years, and I have always liked it here. However, I don't know why my friends and masters are github and stackoverflow. The habit of recording and summarizing from time to time in the learning process is very important.

I'm going to try some other platforms later

• Zhihu: https://www.zhihu.com/people/wynshiter/activities
• https://cn.wordpress.org/
• Brief book:
• https://pages.github.com/

1.6 c++ sdk for machine learning algorithms (more options available)

Dlib is a cross-platform general-purpose library written in modern C++ technology, under the Boost Software licence. It contains machine learning algorithms and tools for creating complex software in C++ to solve real-world problems. It is widely used in various fields in industry and academia, including robotics, embedded devices, mobile phones, and large-scale high-performance computing environments.

The main features are as follows:
1. Complete documentation: Each class and function has detailed documentation, and provides a lot of sample code. If you find that the documentation is unclear or there is no documentation, tell the author, and the author will add it immediately.
2. Portable code: The code conforms to the ISO C++ standard, does not require third-party library support, supports win32, Linux, Mac OS X, Solaris, HPUX, BSDs and POSIX systems
3. Thread support: Provides a simple and portable thread API
4 .Network support: Provide a simple and portable Socket API and a simple Http server
5. Graphical user interface: Provide thread-safe GUI API
6. Numerical algorithm: matrix, large integer, random number operation, etc.
7. Machine learning algorithm:
8. Graphical model algorithm:
9. Image processing: support reading and writing Windows BMP files, different types of color conversion
10. Data compression and integrity algorithms: CRC32, Md5, different forms of PPM algorithms
11. Test: thread-safe log classes and Modular unit testing framework and various test assert support
12. General tools: XML parsing, memory management, type-safe big/little endian conversions, serialization support, and container classes

Reference page :

The five most awesome machine learning projects of 2017
https://www.kdnuggets.com/2017/01/five-machine-learning-projects-cant-overlook-january.html

35 Most Awesome Machine Learning Projects
https://mp.weixin.qq.com/s/zBaOHSMqC7v7dML9AWPLiA

Use dlib's python interface to implement face swap
http://python.jobbole.com/82546/

1.7 Visualize the whole process of machine learning

1. Neural network training: http://playground.tensorflow.org

We choose a complex example where the data is non-linearly sliced

2. Two-layer simple neural network demonstrates nonlinear segmentation
http://cs.stanford.edu/people/karpathy/convnetjs//demo/classify2d.html

3. Neural network for handwritten character recognition
http://terencebroad.com/convnetvis/vis.html

---

2. Transwarp Machine Learning Training

The process of shifting the direction of machine learning is painful. Recently, I spent a week in Shanghai to participate in transwarp's data analyst training. This is the land where I set foot on the magic capital again after a lapse of 7 years. The last time I came here was seven years ago to see the Shanghai World Expo. The difference is that the trip to Shanghai here started from Beijing. I was on the first Fuxing train. Since many foreigners were taking pictures in novels, I passed by the great rivers and mountains of the motherland at a rapid speed, which made me amazed at the speed of development of the motherland with a sense of national pride. In 2010, it would be very difficult to come to Shanghai, especially to buy a sleeper, which is even more difficult.

I am very grateful for the valuable training opportunities given by the leaders of the unit. To be honest, I have not learned the content of machine learning comprehensively and systematically before. But the most important thing is not to know: how to combine machine learning and business in real, industrial level, business. This training basically gave me the answer. transwarp by

I recommend its machine learning product sophon, which supports drag and drop, which allows me to intuitively feel the use of the entire machine learning tool platform and the modeling routines of the machine learning model. There are two things that I haven't paid much attention to before:

1. Feature engineering, normalization, string indexing
2. Evaluation metrics, roc, sum of variance, etc.

2.1 Algorithms of Machine Learning

2.2 How to measure whether a business needs machine learning?

1. Are machine learning algorithms applicable to the business problem?
2. How to choose a model
3. Design development rhythm
4. Final product inspection

2.3 Complete data mining modeling process

2.4 Feature Engineering

Feature engineering is the decisive factor in machine learning and the key to machine learning success

"Data and features determine the upper limit of machine learning, and models and algorithms are only approaching this upper limit."
Throughout Kaggle, KDD, Ali Tianchi and other domestic and foreign competitions, the champion of each competition
did not use very deep Most of the algorithms have done a good job in feature engineering,
and then using some common algorithms, such as Linear Regression, can get excellent
performance .

domain specific knowledge,

Recently, I also saw some good articles on the public account, and I will share them after I have gathered and understood them.

2.5 Visual Drag and Drop Machine Learning Products

I feel that the following two products are somewhat similar

KNIME https://www.knime.com/

One of the biggest challenges for beginners in machine learning and data science is the need to learn too much at the same time, especially if you don't know how to code. You need to quickly adapt to linear algebra, statistics, and other math concepts and learn how to code them, which can be a bit overwhelming for new users.

If you don't have a background in coding and find it difficult to learn, this is where you can use a GUI-driven tool to learn data science. When you're just starting out, focus on learning the actual project. Once you get used to the basic concepts, you can slowly learn how to code later.

In today's article, I will introduce a GUI based tool: KNIME

sophon

Transwarp Sophon also helps data engineers develop data mining applications through Transwarp Sophon. Sophon provides a visual interface tool Midas for creating models. Users only need to drag and drop data source objects and operators to complete the model design, and then train or predict the model on the TDH cluster.

In addition, Sophon also integrates the deep learning framework Tensorflow, which enables users to generate various neural network models by dragging and dropping, flexibly adjust parameters and train, and combine big data and artificial intelligence to promote business innovation.

3. Classic Cases - Survival Prediction of Titanic Passengers, Housing Price Prediction

http://blog.csdn.net/sinat_32547403/article/details/71269804
http://blog.csdn.net/ZengHaihong/article/details/53199559

4. Excellent open source machine learning library

28 Most Popular Open Source Machine Learning Projects on GitHub
https://yq.aliyun.com/articles/30794

15 Open Source Top AI Tools
http://blog.jobbole.com/106447/

5. How ordinary programmers learn machine learning

https://www.zhihu.com/question/51039416

To be continued. . . .