Sequential feature selection for classification

Authors:
Thomas Rückstieß;Christian Osendorfer;Patrick van der Smagt
Affiliations:
Technische Universität München, Garching, Germany;Technische Universität München, Garching, Germany;German Aerospace Center / DLR, Wessling, Germany
Venue:
AI'11 Proceedings of the 24th international conference on Advances in Artificial Intelligence
Year:
2011

Citing 8
Cited 1

An efficient gradient-based algorithm for on-line training of recurrent network trajectories

Neural Computation
Locally Weighted Projection Regression: Incremental Real Time Learning in High Dimensional Space

ICML '00 Proceedings of the Seventeenth International Conference on Machine Learning
Tree-Based Batch Mode Reinforcement Learning

The Journal of Machine Learning Research
Long Short-Term Memory

Neural Computation
Handling Missing Values when Applying Classification Models

The Journal of Machine Learning Research
Gaussian process dynamic programming

Neurocomputing
Attention control with reinforcement learning for face recognition under partial occlusion

Machine Vision and Applications
Neural fitted q iteration – first experiences with a data efficient neural reinforcement learning method

ECML'05 Proceedings of the 16th European conference on Machine Learning

Feature selection filter for classification of power system operating states

Computers & Mathematics with Applications

Quantified Score

Hi-index	0.00

Visualization

Abstract

In most real-world information processing problems, data is not a free resource; its acquisition is rather time-consuming and/or expensive. We investigate how these two factors can be included in supervised classification tasks by deriving classification as a sequential decision process and making it accessible to Reinforcement Learning. Our method performs a sequential feature selection that learns which features are most informative at each timestep, choosing the next feature depending on the already selected features and the internal belief of the classifier. Experiments on a handwritten digits classification task show significant reduction in required data for correct classification, while a medical diabetes prediction task illustrates variable feature cost minimization as a further property of our algorithm.