Incorporating Prior Knowledge on Features into Learning

Eyal Krupka, Naftali Tishby
; Proceedings of the Eleventh International Conference on Artificial Intelligence and Statistics, PMLR 2:227-234, 2007.

Abstract

In the standard formulation of supervised learning the input is represented as a vector of features. However, in most real-life problems, we also have additional information about each of the features. This information can be represented as a set of properties, referred to as meta-features. For instance, in an image recognition task, where the features are pixels, the meta-features can be the (x, y) position of each pixel. We propose a new learning framework that incorporates meta- features. In this framework we assume that a weight is assigned to each feature, as in linear discrimination, and we use the meta-features to define a prior on the weights. This prior is based on a Gaussian process and the weights are assumed to be a smooth function of the meta-features. Using this framework we derive a practical algorithm that improves gen- eralization by using meta-features and discuss the theoretical advantages of incorporating them into the learning. We apply our framework to design a new kernel for hand-written digit recognition. We obtain higher accuracy with lower computational complexity in the primal representation. Finally, we discuss the applicability of this framework to biological neural networks.

Cite this Paper


BibTeX
@InProceedings{pmlr-v2-krupka07a, title = {Incorporating Prior Knowledge on Features into Learning}, author = {Eyal Krupka and Naftali Tishby}, booktitle = {Proceedings of the Eleventh International Conference on Artificial Intelligence and Statistics}, pages = {227--234}, year = {2007}, editor = {Marina Meila and Xiaotong Shen}, volume = {2}, series = {Proceedings of Machine Learning Research}, address = {San Juan, Puerto Rico}, month = {21--24 Mar}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v2/krupka07a/krupka07a.pdf}, url = {http://proceedings.mlr.press/v2/krupka07a.html}, abstract = {In the standard formulation of supervised learning the input is represented as a vector of features. However, in most real-life problems, we also have additional information about each of the features. This information can be represented as a set of properties, referred to as meta-features. For instance, in an image recognition task, where the features are pixels, the meta-features can be the (x, y) position of each pixel. We propose a new learning framework that incorporates meta- features. In this framework we assume that a weight is assigned to each feature, as in linear discrimination, and we use the meta-features to define a prior on the weights. This prior is based on a Gaussian process and the weights are assumed to be a smooth function of the meta-features. Using this framework we derive a practical algorithm that improves gen- eralization by using meta-features and discuss the theoretical advantages of incorporating them into the learning. We apply our framework to design a new kernel for hand-written digit recognition. We obtain higher accuracy with lower computational complexity in the primal representation. Finally, we discuss the applicability of this framework to biological neural networks.} }
Endnote
%0 Conference Paper %T Incorporating Prior Knowledge on Features into Learning %A Eyal Krupka %A Naftali Tishby %B Proceedings of the Eleventh International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2007 %E Marina Meila %E Xiaotong Shen %F pmlr-v2-krupka07a %I PMLR %J Proceedings of Machine Learning Research %P 227--234 %U http://proceedings.mlr.press %V 2 %W PMLR %X In the standard formulation of supervised learning the input is represented as a vector of features. However, in most real-life problems, we also have additional information about each of the features. This information can be represented as a set of properties, referred to as meta-features. For instance, in an image recognition task, where the features are pixels, the meta-features can be the (x, y) position of each pixel. We propose a new learning framework that incorporates meta- features. In this framework we assume that a weight is assigned to each feature, as in linear discrimination, and we use the meta-features to define a prior on the weights. This prior is based on a Gaussian process and the weights are assumed to be a smooth function of the meta-features. Using this framework we derive a practical algorithm that improves gen- eralization by using meta-features and discuss the theoretical advantages of incorporating them into the learning. We apply our framework to design a new kernel for hand-written digit recognition. We obtain higher accuracy with lower computational complexity in the primal representation. Finally, we discuss the applicability of this framework to biological neural networks.
RIS
TY - CPAPER TI - Incorporating Prior Knowledge on Features into Learning AU - Eyal Krupka AU - Naftali Tishby BT - Proceedings of the Eleventh International Conference on Artificial Intelligence and Statistics PY - 2007/03/11 DA - 2007/03/11 ED - Marina Meila ED - Xiaotong Shen ID - pmlr-v2-krupka07a PB - PMLR SP - 227 DP - PMLR EP - 234 L1 - http://proceedings.mlr.press/v2/krupka07a/krupka07a.pdf UR - http://proceedings.mlr.press/v2/krupka07a.html AB - In the standard formulation of supervised learning the input is represented as a vector of features. However, in most real-life problems, we also have additional information about each of the features. This information can be represented as a set of properties, referred to as meta-features. For instance, in an image recognition task, where the features are pixels, the meta-features can be the (x, y) position of each pixel. We propose a new learning framework that incorporates meta- features. In this framework we assume that a weight is assigned to each feature, as in linear discrimination, and we use the meta-features to define a prior on the weights. This prior is based on a Gaussian process and the weights are assumed to be a smooth function of the meta-features. Using this framework we derive a practical algorithm that improves gen- eralization by using meta-features and discuss the theoretical advantages of incorporating them into the learning. We apply our framework to design a new kernel for hand-written digit recognition. We obtain higher accuracy with lower computational complexity in the primal representation. Finally, we discuss the applicability of this framework to biological neural networks. ER -
APA
Krupka, E. & Tishby, N.. (2007). Incorporating Prior Knowledge on Features into Learning. Proceedings of the Eleventh International Conference on Artificial Intelligence and Statistics, in PMLR 2:227-234

Related Material