A gradient based technique for generating sparse representation in function approximation

S. Vijayakumar; S. Wu

doi:10.1109/ICONIP.1999.844006

A gradient based technique for generating sparse representation in function approximation

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

We provide an RKHS based inverse problem formulation for analytically deriving the optimal function approximation when probabilistic information about the underlying regression is available in terms of the associated correlation functions as used by Poggio and Girosi (1998) and Peney and Atick (1996). On the lines of Poggio and Girosi, we show that this solution can be sparsified using principles of SVM and provide an implementation of this sparsification using a novel, conceptually simple and robust gradient based sequential method instead of the conventional quadratic programming routines

Original language	English
Title of host publication	Neural Information Processing, 1999. Proceedings. ICONIP'99. 6th International Conference on
Publisher	Institute of Electrical and Electronics Engineers (IEEE)
Pages	314-319
Number of pages	6
Volume	1
ISBN (Print)	0-7803-5871-6
DOIs	https://doi.org/10.1109/ICONIP.1999.844006
Publication status	Published - 1999

Access to Document

10.1109/ICONIP.1999.844006

http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=844006

Cite this

@inproceedings{81713ac96bad4f28abc09faedc2d66b5,

title = "A gradient based technique for generating sparse representation in function approximation",

abstract = "We provide an RKHS based inverse problem formulation for analytically deriving the optimal function approximation when probabilistic information about the underlying regression is available in terms of the associated correlation functions as used by Poggio and Girosi (1998) and Peney and Atick (1996). On the lines of Poggio and Girosi, we show that this solution can be sparsified using principles of SVM and provide an implementation of this sparsification using a novel, conceptually simple and robust gradient based sequential method instead of the conventional quadratic programming routines",

author = "S. Vijayakumar and S. Wu",

year = "1999",

doi = "10.1109/ICONIP.1999.844006",

language = "English",

isbn = "0-7803-5871-6",

volume = "1",

pages = "314--319",

booktitle = "Neural Information Processing, 1999. Proceedings. ICONIP'99. 6th International Conference on",

publisher = "Institute of Electrical and Electronics Engineers (IEEE)",

address = "United States",

}

A gradient based technique for generating sparse representation in function approximation. / Vijayakumar, S.; Wu, S.

Neural Information Processing, 1999. Proceedings. ICONIP'99. 6th International Conference on. Vol. 1 Institute of Electrical and Electronics Engineers (IEEE), 1999. p. 314-319.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - A gradient based technique for generating sparse representation in function approximation

AU - Vijayakumar, S.

AU - Wu, S.

PY - 1999

Y1 - 1999

N2 - We provide an RKHS based inverse problem formulation for analytically deriving the optimal function approximation when probabilistic information about the underlying regression is available in terms of the associated correlation functions as used by Poggio and Girosi (1998) and Peney and Atick (1996). On the lines of Poggio and Girosi, we show that this solution can be sparsified using principles of SVM and provide an implementation of this sparsification using a novel, conceptually simple and robust gradient based sequential method instead of the conventional quadratic programming routines

AB - We provide an RKHS based inverse problem formulation for analytically deriving the optimal function approximation when probabilistic information about the underlying regression is available in terms of the associated correlation functions as used by Poggio and Girosi (1998) and Peney and Atick (1996). On the lines of Poggio and Girosi, we show that this solution can be sparsified using principles of SVM and provide an implementation of this sparsification using a novel, conceptually simple and robust gradient based sequential method instead of the conventional quadratic programming routines

U2 - 10.1109/ICONIP.1999.844006

DO - 10.1109/ICONIP.1999.844006

M3 - Conference contribution

SN - 0-7803-5871-6

VL - 1

SP - 314

EP - 319

BT - Neural Information Processing, 1999. Proceedings. ICONIP'99. 6th International Conference on

PB - Institute of Electrical and Electronics Engineers (IEEE)

ER -

A gradient based technique for generating sparse representation in function approximation

Abstract

Access to Document

Fingerprint

Cite this