Recurrent Neural Networks for Adaptive Feature Acquisition

被引：11

作者：

Contardo, Gabriella ^{[1
]}

Denoyer, Ludovic ^{[1
]}

Artieres, Thierry ^{[2
]}

机构：

[1] Univ Paris 06, Sorbonne Univ, LIP6, UMR 7606, F-75005 Paris, France

[2] Aix Marseille Univ, CNRS, LIF, Cent Marseille, Marseille, France

来源：

NEURAL INFORMATION PROCESSING, ICONIP 2016, PT III | 2016年 / 9949卷

关键词：

COST;

D O I：

10.1007/978-3-319-46675-0_65

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

We propose to tackle the cost-sensitive learning problem, where each feature is associated to a particular acquisition cost. We propose a new model with the following key properties: (i) it acquires features in an adaptive way, (ii) features can be acquired per block (several at a time) so that this model can deal with high dimensional data, and (iii) it relies on representation-learning ideas. The effectiveness of this approach is demonstrated on several experiments considering a variety of datasets and with different cost settings.

引用

页码：591 / 599

页数：9