Machine Learning

Machine Learning is a very popular approach in artificial intelligence. NeoNeuro Data Mining is an application which learns from examples like a child. 

NeoNeuro Data Mining is easy free and convenient way to understand and use machine learning technologies.

1. Main idea of NeoNeuro Machine Learning is universality.
2. Universality is when everything works over ONE button. Copy data into spreadsheet and press Calculate – that’s all!
3. By default Machine Learning opens the example where PC learns elementary math operations. We recommend to teach computer from scratch to add and divide in order to understand how the programs works.
4. Press ChessBoard and teach the Machine Learning to play chess moves. You will see how computer makes mistakes in the beginning and how it gradually learns to move correctly. Artificial intelligence of NeoNeuro Machine Learning is like a human one in this aspect.
5. “Humanity” of application can also be seen in work with statistical databases. Unlike classical data mining and machine learning programs, NeoNeuro Machine Learning can answer “I don’t know” or give optional answers, exactly like a human expert.
6. Universality of Machine Learning is
   1. Unified method of work with any data: statistical databases, learning arithmetic, solving geometry problems like teaching chess moves or robot movements.
   2. Easy method to add data. Just copy them from a spreadsheet or from HTML page and paste into NeoNeuro Machine Learning. Application will automatically understand parameters types, which can be also set by hand.
   3. Input data for test is the same as for learning, here you can also use copy and paste operations.
   4. Machine Learning gives detailed and clear report. For statistical data 10% of learning data is not used for learning, it is used for test. In the report, you will see confusion matrix and pie diagram of answers distribution.
7. Settings. By default Machine Learning is designed in a way where you can comfortably work with any data without special settings. In addition, it is possible to change:
   1. Attributes types
   2. Dimensions of attributes. For instance, salary and monthly credit fee both have  “Money” dimension, while age and number of years to pay credit off have  “Time” dimension.
   3. In Machine Learning you can change the Confidence parameter. Better confidence gives more “I don’t know” answers, lower confidence accepts more variants to give the exact answer. This is like a human confidence.
8. Today Machine Learning is the only known application which can learn chess moves. This skill appeared due to usage of the geometry dimensions: X and Y. Machine Learning can be used for teaching robots to move. In Machine Learning you can teach a Monkey to come to the banana, the same algorithm can be used in order to teach robot to make complicated manipulations from cleaning and house building to making surgery. Moreover, the learning process is similar to the way how a human learns. Robot is shown how to make the particular work and is told what mistakes he makes, in order to avoid them in the future.
9. Work with geometry is not the only unique feature of Machine Learning. Algorithm gives better results in work with statistical datasets than most of classical data mining algorithms. It is also connected with the feature of Machine Learning to give answers “I don’t know” or give optional answer if the confidence in exact answer is not enough.  Error rate in Machine Learning is usually lower than in methods like logistical regression, neural nets and other data mining algorithms.
10. In future versions of Machine Learning we are planning the following:
    1. API for development in VC++, C# and Delphi
    2. Translation to many languages.