A pipeline approach in identifying important input features from neural networks

Yuyu He, Chih Lai, Dalma Martinovic-Weigelt, Zezheng Long

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Neural networks are well-known for their powerful capability in producing high prediction accuracy. However, due to the non-linear calculations in the network, it is very difficult for users to understand which input features are important in leading to final predictions. In this study, we propose a two-step pipeline approach that uses two sets of linear models to estimates feature importance in the input dataset X that leads to the class prediction specified in Y. More specifically, the first linear regression model derives the feature importance in X in explaining the Z-code that was extracted from any hidden layer of a trained neural network. The second linear classification model captures the importance in the Z- code in predicting the target class Y. We then combine the first X to Z importance with the second Z to Y importance together to approximate the non-linear importance from X to Y. The experiments conducted in this study also show that our method is sound and stable in selecting the truly important features from input datasets regardless how a neural network was constructed with different parameters such as activation functions or the number of hidden layers.

Original languageEnglish (US)
Title of host publication2019 14th Annual Conference System of Systems Engineering, SoSE 2019
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages25-30
Number of pages6
ISBN (Electronic)9781728104577
DOIs
StatePublished - May 2019
Externally publishedYes
Event14th Annual Conference System of Systems Engineering, SoSE 2019 - Anchorage, United States
Duration: May 19 2019May 22 2019

Publication series

Name2019 14th Annual Conference System of Systems Engineering, SoSE 2019

Conference

Conference14th Annual Conference System of Systems Engineering, SoSE 2019
Country/TerritoryUnited States
CityAnchorage
Period5/19/195/22/19

Bibliographical note

Publisher Copyright:
© 2019 IEEE.

Keywords

  • Autoencoder
  • Hidden layer
  • Linear regression
  • Logistic regression
  • Neural network

Fingerprint

Dive into the research topics of 'A pipeline approach in identifying important input features from neural networks'. Together they form a unique fingerprint.

Cite this