Definition of machine learning and supervised learning and unsupervised learning

Machine learning Definition

Machine learning is a branch of AI that aims to give machines a new ability. (specialized in how computers simulate or implement human learning behaviors in order to acquire new knowledge or skills and reorganize existing knowledge structures to continually improve their performance.) machine learning is widely used, such as large-scale data mining (web-click Data, medical records, etc.), drones, cars, hand writers, most natural language processing tasks, computer vision, recommender systems, etc.

Machine learning has many definitions and is well known for the following two articles:

Arthursamuel (1959): Machine Learning:field of study, gives computers theability to learn without being explicitly PR Ogrammed.

Tommitchell (1998): Well-posed Learning problem:a computer program was said Tolearn from experience E with respect to SOM The e task T and some performance measurep, if its performance on T, as measured by P, improves with experience E.

Example: For a spam identification problem, classifying messages as spam or non-spam is task T, to see which messages are marked as spam and which are marked as non-spam is experience E, the number of correctly identified spam or non-spam messages or ratios is the benchmark indicator p.

Supervised learning

Training samples with concept marks (classifications) are learned to mark (classify) predictions of data outside the training sample set as much as possible. Here, all the tags (categories) are known. Therefore, the ambiguity of the training sample is low.

Supervised learning is the most common technique for training neural networks and decision trees. These two technologies (neural networks and decision trees) are highly dependent on the information given by the pre-determined classification system.

House price Prediction-Regression (Regression): Predicting continuous output values (price)

breast cancer (benign, malignant) prediction problems - category (classification): predicting discrete output values (0, 1)

It can be processed even with an infinite number of features (support vector machines).

Classification and regression are the contents of supervised learning.

Unsupervised learning

Training samples without concept marks (classifications) are studied to discover the structural knowledge in the training sample set. Here, all the tags (categories) are unknown. Therefore, the training sample is of high ambiguity.

The common unsupervised learning algorithms are clustering.

The above describes supervised learning. Recalling the data set at the time, the table shows that each piece of data in the data set has been labeled negative or positive, i.e. benign or malignant. So, for each piece of data in the supervised learning, we already know clearly that the correct answer to the training set is benign or malignant.

In unsupervised learning, we know the data. It looks a little different, unlike the look of the supervised learning data, i.e. unsupervised learning without any tags or the same tags. for data sets, unsupervised learning can determine that there are two different clusters of data. Unsupervised learning algorithms may divide these data into two different clusters. So called the clustering algorithm, it can be used in many places.

Unsupervised learning has a lot of applications. It is used to organize clusters of large computers. The second application is the analysis of social networks. There is also market segmentation. Many companies have large databases that store consumer information. So, you can retrieve these customer data sets, automatically discover the market classification, and automatically divide the customer into different market segments so that you can automatically and effectively sell or sell together in different segments of the market. Finally, unsupervised learning can also be used for astronomical data analysis, which gives surprising, interesting and useful theories that explain how galaxies were born. These are all examples of clustering, and clustering is just one of unsupervised learning .

Note: This article is a study note from Professor Andrew Ng's "Machine vision" course.

Reprint Please specify source: http://blog.csdn.net/lsh_2013/article/details/47342641

Copyright NOTICE: This article for Bo Master original article, without Bo Master permission not reproduced.

Definition of machine learning and supervised learning and unsupervised learning