Understand that the development of supervised, unsupervised, and semi-supervised feature selection are depend on the class labeled in a data set, either labeled, unlabeled or partially labeled. But some feature selection algorithm like Sequential Forward selection (SFS), Spectral feature selection (SPEC), PSO, GA and etc, have used supervised machine learning (SVM, KNN, neural network) for classification task. So is those feature selection algorithm with supervised machine learning still considers as unsupervised / semi-supervised FS algorithm?
Can I assume :
Feature selection + supervised machine learning (classification method) = Supervised Feature selection ?
Feature selection + unsupervised ML (clustering method) = unsupervised Feature selection ?
In the literature, feature selection methods are typically partitioned into "filter" methods (i.e., used as a preprocessing step) and "wrapper" methods (i.e., embedded into a learning procedure) [1-2]. In the former case, you can have supervised and unsupervised feature selection independently of any classifier: as an example, correlation-based feature selection is supervised, while principal component analysis is not.
[1] Kohavi, Ron, and George H. John. "Wrappers for feature subset selection." Artificial intelligence 97.1 (1997): 273-324.
[2] Guyon, Isabelle, and André Elisseeff. "An introduction to variable and feature selection." The Journal of Machine Learning Research 3 (2003): 1157-1182.
In the literature, feature selection methods are typically partitioned into "filter" methods (i.e., used as a preprocessing step) and "wrapper" methods (i.e., embedded into a learning procedure) [1-2]. In the former case, you can have supervised and unsupervised feature selection independently of any classifier: as an example, correlation-based feature selection is supervised, while principal component analysis is not.
[1] Kohavi, Ron, and George H. John. "Wrappers for feature subset selection." Artificial intelligence 97.1 (1997): 273-324.
[2] Guyon, Isabelle, and André Elisseeff. "An introduction to variable and feature selection." The Journal of Machine Learning Research 3 (2003): 1157-1182.
JC: I think Simone's response clarifies that your assumptions don't generally hold, right?
- Badges
- Science topic
- Similar topics
- Computer Science
- Data Mining