I'm working on feature selection methods. I found feature reduction in rough sets theory a concept similar to feature selection in machine learning and data processing tasks. However it seems that they have some principal differences. Can anybody explain if I can use them in place of each other?
There is no clear boundary. In all cases, we are able to use the same methods. However, when it comes to the application of neural networks and traditional methods, the difference is obvious. In the first case, due to training, we associate the elements of the signals, their characteristics and other features. Then, we use them for recognition and selection. Using traditional methods, we are hard-coded rules. As regards the theory of sets in the space of breeding, including the use of neural networks, see the article " Cover T. M. Geometrical and Statistical Properties of Systems of Linear Inequalities with Applications in Pattern recognition // IEEE Transactions on Electronic Computers, June, 1965".
In machine learning, feature reduction includes feature selection and feature extraction. Feature reduction aims to reduce the number of features, so that the problem can be computed by the machine. Both feature selection and feature extraction can reduce the feature number. Moreover, feature selection can improve the results by reducing the interference brought by some useless features, while feature extraction can provide you a less number of new features extracted from those original features.
Dear friend
Greetings.
I would like to add some spice to the above answers provided by above researchers.
Feature reduction and selection are used in the literature interchangingly to indicate the same point.
To select a specific set of Features from a set of a given features, it is required to reduce the irrelevant features by eliminating them by using some methods. The method may be wrapper based or filter based or combination of them. You may look at my publications as I have used a variety of feature selection methods.
I hope this helps you.
Best regards
Dr.Indrajit Mandal, Ph.D.
All this kind of techniques come under pre-processing of date.
But each technique is having its bit difference
Feature selection is to find the best feature select for your application.
Feature reduction is to reduce or dimentionlaity reduction for the selected features.
Dear Mr.Sadegh
In feature selection we have two part feature subset selection(Wrapping method) and filtering method.
In wrapping we select attributes that are important and discard others with forward and backward method
On the other hand in filtering we try to find relevance feature and one of this that calculate with variance is PCA method that extract some new feature that are fewer than original feature
All and all, I think feature selection is our aim that maybe led to feature reduction
- Badges
- Science topic