Why 10,402,685 patent is so important, massive history just massive
United States Patent 10,402,685
Guyon , et al. September 3, 2019
Recursive feature elimination method using support vector machines
Abstract
Identification of a determinative subset of features from within a group of features is
performed by training a support vector machine
using training samples with class
labels to determine a value of each feature,
where features are removed based on their
the value. One or more features having the
smallest values are removed and an updated
kernel matrix is generated using the remaining
features. The process is repeated until a
predetermined number of features remain which
are capable of accurately separating the data
into different classes. In some embodiments,
features are eliminated by a ranking criterion
based on a Lagrange multiplier corresponding to
each training sample.
Inventors: Guyon; Isabelle (Berkeley, CA), Weston; Jason Aaron Edward (New York, NY)
Applicant:
Name City State Country Type
Guyon; Isabelle
Weston; Jason Aaron Edward
Berkeley
New York
CA
NY
US
US
Assignee: HEALTH DISCOVERY CORPORATION (Atlanta, GA)
Family ID: 43781396
Appl. No.: 12/944,197
Filed: November 11, 2010
Prior Publication Data
Document Identifier Publication Date
US 20110106735 A1 May 5, 2011
Related U.S. Patent Documents
Application Number Filing Date Patent Number Issue Date
12890705 Sep 26, 2010 7970718
11928784 Sep 28, 2010 7805388
10494876 Jan 6, 2009 7475048
PCT/US02/35576 Nov 7, 2002
PCT/US02/16012 Jan 8, 2008 7318051
10057849 Oct 3, 2006 7117188
09633410 Apr 19, 2005 6882990
60347562 Nov 7, 2001
60292133 May 18, 2001
60292221 May 18, 2001
60332021 Nov 21, 2001
60161806 Oct 27, 1999
60168703 Dec 2, 1999
60184596 Feb 24, 2000
60191219 Mar 22, 2000
Current U.S. Class: 1/1
Current CPC Class: G06N 20/10 (20190101); G06N 20/00 (20190101); G06K 9/6231 (20130101); G16B 40/00 (20190201); G16B 25/00 (20190201)
Current International Class: G06N 20/00 (20190101); G06K 9/62 (20060101); G16B 40/00 (20190101); G06N 20/10 (20190101); G16B 25/00 (20190101)
References Cited [Referenced By]
U.S. Patent Documents
4881178 November 1989 Holland
5138694 August 1992 Hamilton
5255347 October 1993 Matsuba et al.
5649068 July 1997 Boser et al.
5809144 September 1998 Sirbu
5819246 October 1998 Ashida et al.
5950146 September 1999 Vapnik et al.
6087098 July 2000 McKiernan et al.
6112195 August 2000 Burges
6128608 October 2000 Barnhill
6134344 October 2000 Burges
6157921 December 2000 Barnhill
6161130 December 2000 Horvitz et al.
6187549 February 2001 Schmidt et al.
6251586 June 2001 Mulshine et al.
6427141 July 2002 Barnhill
6532305 March 2003 Hammen
6647341 November 2003 Golub et al.
6658395 December 2003 Barnhill
6714925 March 2004 Barnhill et al.
6760715 July 2004 Barnhill et al.
6789069 September 2004 Barnhill et al.
6882990 April 2005 Barnhill et al.
6944602 September 2005 Cristianini
6996549 February 2006 Zhang et al.
7047137 May 2006 Kasif et al.
7117188 October 2006 Guyon et al.
7299213 November 2007 Cristianini
7318051 January 2008 Weston et al.
7383237 June 2008 Zhang et al.
7444308 October 2008 Guyon et al.
7475048 January 2009 Weston et al.
7542947 June 2009 Guyon et al.
7542959 June 2009 Barnhill et al.
7617163 November 2009 Ben-Hur et al.
7624074 November 2009 Weston et al.
7676442 March 2010 Ben-Hur et al.
7685077 March 2010 Li et al.
7797257 September 2010 Barnhill et al.
7805388 September 2010 Weston et al.
2003/0036081 February 2003 Adorjan
2003/0086593 May 2003 Liu et al.
2004/0005547 January 2004 Boess et al.
2004/0102905 May 2004 Adorjan
2005/0069936 March 2005 Diamond et al.
2005/0181386 August 2005 Diamond et al.
2006/0063156 March 2006 Willman et al.
2006/0074821 April 2006 Cristianini
2010/0256988 October 2010 Barnhill et al.
Other References
United States Patent 10,402,685
Guyon , et al. September 3, 2019
Recursive feature elimination method using support vector machines
Abstract
Identification of a determinative subset of features from within a group of features is
performed by training a support vector machine
using training samples with class
labels to determine a value of each feature,
where features are removed based on their
the value. One or more features having the
smallest values are removed and an updated
kernel matrix is generated using the remaining
features. The process is repeated until a
predetermined number of features remain which
are capable of accurately separating the data
into different classes. In some embodiments,
features are eliminated by a ranking criterion
based on a Lagrange multiplier corresponding to
each training sample.
Inventors: Guyon; Isabelle (Berkeley, CA), Weston; Jason Aaron Edward (New York, NY)
Applicant:
Name City State Country Type
Guyon; Isabelle
Weston; Jason Aaron Edward
Berkeley
New York
CA
NY
US
US
Assignee: HEALTH DISCOVERY CORPORATION (Atlanta, GA)
Family ID: 43781396
Appl. No.: 12/944,197
Filed: November 11, 2010
Prior Publication Data
Document Identifier Publication Date
US 20110106735 A1 May 5, 2011
Related U.S. Patent Documents
Application Number Filing Date Patent Number Issue Date
12890705 Sep 26, 2010 7970718
11928784 Sep 28, 2010 7805388
10494876 Jan 6, 2009 7475048
PCT/US02/35576 Nov 7, 2002
PCT/US02/16012 Jan 8, 2008 7318051
10057849 Oct 3, 2006 7117188
09633410 Apr 19, 2005 6882990
60347562 Nov 7, 2001
60292133 May 18, 2001
60292221 May 18, 2001
60332021 Nov 21, 2001
60161806 Oct 27, 1999
60168703 Dec 2, 1999
60184596 Feb 24, 2000
60191219 Mar 22, 2000
Current U.S. Class: 1/1
Current CPC Class: G06N 20/10 (20190101); G06N 20/00 (20190101); G06K 9/6231 (20130101); G16B 40/00 (20190201); G16B 25/00 (20190201)
Current International Class: G06N 20/00 (20190101); G06K 9/62 (20060101); G16B 40/00 (20190101); G06N 20/10 (20190101); G16B 25/00 (20190101)
References Cited [Referenced By]
U.S. Patent Documents
4881178 November 1989 Holland
5138694 August 1992 Hamilton
5255347 October 1993 Matsuba et al.
5649068 July 1997 Boser et al.
5809144 September 1998 Sirbu
5819246 October 1998 Ashida et al.
5950146 September 1999 Vapnik et al.
6087098 July 2000 McKiernan et al.
6112195 August 2000 Burges
6128608 October 2000 Barnhill
6134344 October 2000 Burges
6157921 December 2000 Barnhill
6161130 December 2000 Horvitz et al.
6187549 February 2001 Schmidt et al.
6251586 June 2001 Mulshine et al.
6427141 July 2002 Barnhill
6532305 March 2003 Hammen
6647341 November 2003 Golub et al.
6658395 December 2003 Barnhill
6714925 March 2004 Barnhill et al.
6760715 July 2004 Barnhill et al.
6789069 September 2004 Barnhill et al.
6882990 April 2005 Barnhill et al.
6944602 September 2005 Cristianini
6996549 February 2006 Zhang et al.
7047137 May 2006 Kasif et al.
7117188 October 2006 Guyon et al.
7299213 November 2007 Cristianini
7318051 January 2008 Weston et al.
7383237 June 2008 Zhang et al.
7444308 October 2008 Guyon et al.
7475048 January 2009 Weston et al.
7542947 June 2009 Guyon et al.
7542959 June 2009 Barnhill et al.
7617163 November 2009 Ben-Hur et al.
7624074 November 2009 Weston et al.
7676442 March 2010 Ben-Hur et al.
7685077 March 2010 Li et al.
7797257 September 2010 Barnhill et al.
7805388 September 2010 Weston et al.
2003/0036081 February 2003 Adorjan
2003/0086593 May 2003 Liu et al.
2004/0005547 January 2004 Boess et al.
2004/0102905 May 2004 Adorjan
2005/0069936 March 2005 Diamond et al.
2005/0181386 August 2005 Diamond et al.
2006/0063156 March 2006 Willman et al.
2006/0074821 April 2006 Cristianini
2010/0256988 October 2010 Barnhill et al.
Other References
