Local Feature Selection Using Gaussian Process Regression. Pichara, K. & Soto, A. Intelligent Data Analysis (IDA), 2014. Paper abstract bibtex 22 downloads Most feature selection algorithms determine a global subset of features, where all data instances are projected in order to improve classification accuracy. An attractive alternative solution is to adaptively find a local subset of features for each data instance, such that, the classification of each instance is performed according to its own selective subspace. This paper presents a novel application of Gaussian Processes that improves classification performance by learning discriminative local subsets of features for each instance in a dataset. Gaussian Processes are used to build for each available feature a function that estimates the discriminative power of the feature over all the input space. Using these functions, we are able to determine a discriminative subspace for each possible instance by locally joining the features that present the highest levels of discriminative power. New instances are then classified by using a K-NN classifier that operates in the local subspaces. Experimental results show that by using local discriminative subspaces, we are able to reach higher levels of accuracy than alternative state-of-the-art feature selection approaches.
@Article{ pichara:etal:2014,
author = {K. Pichara and A. Soto},
title = {Local Feature Selection Using Gaussian Process
Regression},
journal = {Intelligent Data Analysis (IDA)},
volume = {18},
number = {3},
year = {2014},
abstract = {Most feature selection algorithms determine a global
subset of features, where all data instances are projected
in order to improve classification accuracy. An attractive
alternative solution is to adaptively find a local subset
of features for each data instance, such that, the
classification of each instance is performed according to
its own selective subspace. This paper presents a novel
application of Gaussian Processes that improves
classification performance by learning discriminative local
subsets of features for each instance in a dataset.
Gaussian Processes are used to build for each available
feature a function that estimates the discriminative power
of the feature over all the input space. Using these
functions, we are able to determine a discriminative
subspace for each possible instance by locally joining the
features that present the highest levels of discriminative
power. New instances are then classified by using a K-NN
classifier that operates in the local subspaces.
Experimental results show that by using local
discriminative subspaces, we are able to reach higher
levels of accuracy than alternative state-of-the-art
feature selection approaches. },
url = {http://saturno.ing.puc.cl/media/papers_alvaro/Karim-IDA-2014.pdf}
}
Downloads: 22
{"_id":{"_str":"53427a470e946d920a0018de"},"__v":1,"authorIDs":["32ZR23o2BFySHbtQK","3ear6KFZSRqbj6YeT","4Pq6KLaQ8jKGXHZWH","54578d9a2abc8e9f370004f0","546f0c3f5ac8e5e30d0000f0","5e126ca5a4cabfdf01000053","5e158f76f1f31adf01000118","5e16174bf67f7dde010003ad","5e1f631ae8f5ddde010000eb","5e1f7182e8f5ddde010001ff","5e26da3642065ede01000066","5e3acefaf2a00cdf010001c8","5e62c3aecb259cde010000f9","5e65830c6e5f4cf3010000e7","5e666dfc46e828de010002c9","6cMBYieMJhf6Nd58M","6w6sGsxYSK2Quk6yZ","7xDcntrrtC62vkWM5","ARw5ReidxxZii9TTZ","BjzM7QpRCG7uCF7Zf","DQ4JRTTWkvKXtCNCp","GbYBJvxugXMriQwbi","HhRoRmBvwWfD4oLyK","JFk6x26H6LZMoht2n","JvArGGu5qM6EvSCvB","LpqQBhFH3PxepH9KY","MT4TkSGzAp69M3dGt","QFECgvB5v2i4j2Qzs","RKv56Kes3h6FwEa55","Rb9TkQ3KkhGAaNyXq","RdND8NxcJDsyZdkcK","SpKJ5YujbHKZnHc4v","TSRdcx4bbYKqcGbDg","W8ogS2GJa6sQKy26c","WTi3X2fT8dzBN5d8b","WfZbctNQYDBaiYW6n","XZny8xuqwfoxzhBCB","Xk2Q5qedS5MFHvjEW","bbARiTJLYS79ZMFbk","cBxsyeZ37EucQeBYK","cFyFQps7W3Sa2Wope","dGRBfr8zhMmbwK6eP","eRLgwkrEk7T7Lmzmf","fMYSCX8RMZap548vv","g6iKCQCFnJgKYYHaP","h2hTcQYuf2PB3oF8t","h83jBvZYJPJGutQrs","jAtuJBcGhng4Lq2Nd","pMoo2gotJcdDPwfrw","q5Zunk5Y2ruhw5vyq","rzNGhqxkbt2MvGY29","uC8ATA8AfngWpYLBq","uoJ7BKv28Q6TtPmPp","vMiJzqEKCsBxBEa3v","vQE6iTPpjxpuLip2Z","wQDRsDjhgpMJDGxWX","wbNg79jvDpzX9zHLK","wk86BgRiooBjy323E","zCbPxKnQGgDHiHMWn","zf9HENjsAzdWLMDAu"],"author_short":["Pichara, K.","Soto, A."],"bibbaseid":"pichara-soto-localfeatureselectionusinggaussianprocessregression-2014","bibdata":{"bibtype":"article","type":"article","author":[{"firstnames":["K."],"propositions":[],"lastnames":["Pichara"],"suffixes":[]},{"firstnames":["A."],"propositions":[],"lastnames":["Soto"],"suffixes":[]}],"title":"Local Feature Selection Using Gaussian Process Regression","journal":"Intelligent Data Analysis (IDA)","volume":"18","number":"3","year":"2014","abstract":"Most feature selection algorithms determine a global subset of features, where all data instances are projected in order to improve classification accuracy. An attractive alternative solution is to adaptively find a local subset of features for each data instance, such that, the classification of each instance is performed according to its own selective subspace. This paper presents a novel application of Gaussian Processes that improves classification performance by learning discriminative local subsets of features for each instance in a dataset. Gaussian Processes are used to build for each available feature a function that estimates the discriminative power of the feature over all the input space. Using these functions, we are able to determine a discriminative subspace for each possible instance by locally joining the features that present the highest levels of discriminative power. New instances are then classified by using a K-NN classifier that operates in the local subspaces. Experimental results show that by using local discriminative subspaces, we are able to reach higher levels of accuracy than alternative state-of-the-art feature selection approaches. ","url":"http://saturno.ing.puc.cl/media/papers_alvaro/Karim-IDA-2014.pdf","bibtex":"@Article{\t pichara:etal:2014,\n author\t= {K. Pichara and A. Soto},\n title\t\t= {Local Feature Selection Using Gaussian Process\n\t\t Regression},\n journal\t= {Intelligent Data Analysis (IDA)},\n volume\t= {18},\n number\t= {3},\n year\t\t= {2014},\n abstract\t= {Most feature selection algorithms determine a global\n\t\t subset of features, where all data instances are projected\n\t\t in order to improve classification accuracy. An attractive\n\t\t alternative solution is to adaptively find a local subset\n\t\t of features for each data instance, such that, the\n\t\t classification of each instance is performed according to\n\t\t its own selective subspace. This paper presents a novel\n\t\t application of Gaussian Processes that improves\n\t\t classification performance by learning discriminative local\n\t\t subsets of features for each instance in a dataset.\n\t\t Gaussian Processes are used to build for each available\n\t\t feature a function that estimates the discriminative power\n\t\t of the feature over all the input space. Using these\n\t\t functions, we are able to determine a discriminative\n\t\t subspace for each possible instance by locally joining the\n\t\t features that present the highest levels of discriminative\n\t\t power. New instances are then classified by using a K-NN\n\t\t classifier that operates in the local subspaces.\n\t\t Experimental results show that by using local\n\t\t discriminative subspaces, we are able to reach higher\n\t\t levels of accuracy than alternative state-of-the-art\n\t\t feature selection approaches. },\n url\t\t= {http://saturno.ing.puc.cl/media/papers_alvaro/Karim-IDA-2014.pdf}\n}\n\n","author_short":["Pichara, K.","Soto, A."],"key":"pichara:etal:2014","id":"pichara:etal:2014","bibbaseid":"pichara-soto-localfeatureselectionusinggaussianprocessregression-2014","role":"author","urls":{"Paper":"http://saturno.ing.puc.cl/media/papers_alvaro/Karim-IDA-2014.pdf"},"metadata":{"authorlinks":{"soto, a":"https://asoto.ing.puc.cl/publications/"}},"downloads":22},"bibtype":"article","biburl":"https://raw.githubusercontent.com/ialab-puc/ialab.ing.puc.cl/master/pubs.bib","downloads":22,"keywords":[],"search_terms":["local","feature","selection","using","gaussian","process","regression","pichara","soto"],"title":"Local Feature Selection Using Gaussian Process Regression","year":2014,"dataSources":["baKak4mfPDQ4RS73F","3YPRCmmijLqF4qHXd","sg6yZ29Z2xB5xP79R","sj4fjnZAPkEeYdZqL","m8qFBfFbjk9qWjcmJ","QjT2DEZoWmQYxjHXS","FEMiN7g9QC58jhiYi"]}