Machine learning is a comprehensive and applied discipline that can be used to solve problems in various fields such as computer vision/biology/robotics and everyday languages, as a result of research on artificial intelligence, and machine learning is designed to enable computers to have the ability to learn as humans do, because we find that computers have the functions to accomplish, Can not be achieved through a fixed program, such as the computer to recognize handwritten handwriting, so that the computer from different images to identify the cat and people.
This course mainly studies machine learning algorithms, but also needs to use the program to implement these algorithms, it is recommended to use MATLAB or octave programming. In addition, learning this course requires knowledge of mathematical statistics/linear algebra and data structures. Machine learning can be defined as follows:
- The research field that makes the computer have learning ability without explicit program. (Arithur Samuel, 1959)
- For a given task T, the computer program learns from experience E, through experience e, to improve the performance of the task T performed p. (Tom mitchell,1998)
Machine learning can be divided into four main parts:
- Supervised learning
Provides a set of standard answers to the algorithm, to supervise the algorithm for the specific input output, is not the answer we give.
The problem of regression and classification can be attributed to supervised learning.
- Learning theory
Learning theory is theoretically to study how to ensure the performance of the learning algorithm, such as how to ensure the accuracy of the algorithm/how many samples needed
- Unsupervised learning
Unsupervised learning is relative to supervised learning, only the data set is given, and no standard answer is given, which is usually used to discover the structural characteristics of the data set. For example, cluster analysis of data, but do not know in advance what these data belong to the class, that is, the discovery of unknown attributes.
- Intensive Learning
Intensive learning is used in situations where you don't need to make a decision. For example, in the case of diagnosis of benign or malignant tumors (this is a supervised learning problem), your decision gives a conclusion that determines the life and death of a patient. However, you might actually need to make multiple decisions in a row over time. For example, an unmanned helicopter's automatic flight, you make a wrong decision, he may not crash immediately, as long as you make the right decision, can be remedied, only if you have been making the wrong decision, will make it crash. Intensive learning can be applied to the field of robotics
Andrew N.G's machine learning public lessons Note (i): Motivation and application of machine learning
