Editing Feature selection (section)

==Feature selection embedded in learning algorithms==
Some learning algorithms perform feature selection as part of their overall operation. These include:

* {{tmath|l_1}}-regularization techniques, such as sparse regression, LASSO, and {{tmath|l_1}}-SVM
* Regularized trees,<ref name="DengRunger2012" /> e.g. regularized random forest implemented in the RRF package<ref name="RRF" />
* [[Decision tree learning|Decision tree]]<ref>R. Kohavi and G. John, "[https://ai.stanford.edu/~ronnyk/wrappersPrint.pdf Wrappers for feature subset selection]", ''[[Artificial Intelligence (journal)|Artificial intelligence]]'' 97.1-2 (1997): 273-324</ref>
* [[Memetic algorithm]]
* [[Random multinomial logit]] (RMNL)
* [[Autoencoder|Auto-encoding]] networks with a bottleneck-layer
* [[Submodular set function|Submodular]] feature selection<ref>{{cite arXiv|eprint=1102.3975|last1=Das|first1=Abhimanyu|title=Submodular meets Spectral: Greedy Algorithms for Subset Selection, Sparse Approximation and Dictionary Selection|last2=Kempe|first2=David|class=stat.ML|year=2011}}</ref><ref>Liu et al., [http://melodi.ee.washington.edu/~bilmes/mypubs/liu-submodfeature2013-icassp.pdf Submodular feature selection for high-dimensional acoustic score spaces] {{Webarchive|url=https://web.archive.org/web/20151017122628/http://melodi.ee.washington.edu/~bilmes/mypubs/liu-submodfeature2013-icassp.pdf |date=2015-10-17 }}</ref><ref>Zheng et al., [http://papers.nips.cc/paper/5565-deep-convolutional-neural-network-for-image-deconvolution Submodular Attribute Selection for Action Recognition in Video] {{Webarchive|url=https://web.archive.org/web/20151118001059/http://papers.nips.cc/paper/5565-deep-convolutional-neural-network-for-image-deconvolution |date=2015-11-18 }}</ref>
* Local learning based feature selection.<ref>{{cite journal | last1 = Sun | first1 = Y. | last2 = Todorovic | first2 = S. | last3 = Goodison | first3 = S. | year = 2010 |title=Local-Learning-Based Feature Selection for High-Dimensional Data Analysis | journal = [[IEEE Transactions on Pattern Analysis and Machine Intelligence]] | volume = 32 | issue = 9| pages = 1610–1626 | doi = 10.1109/tpami.2009.190 | pmc = 3445441 | pmid = 20634556 }}</ref> Compared with traditional methods, it does not involve any heuristic search, can easily handle multi-class problems, and works for both linear and nonlinear problems. It is also supported by a strong theoretical foundation. Numeric experiments showed that the method can achieve a close-to-optimal solution even when data contains >1M irrelevant features.
* Recommender system based on feature selection.<ref>D.H. Wang, Y.C. Liang, D.Xu, X.Y. Feng, R.C. Guan(2018), "[https://www.sciencedirect.com/science/article/pii/S0950705118302107 A content-based recommender system for computer science publications]", ''[[Knowledge-Based Systems]]'', 157: 1-9</ref> The feature selection methods are introduced into recommender system research.