• search hit 5 of 16
Back to Result List

A Measure of Confidence of Artificial Neural Network Classifiers

  • Confidence in results of an Artificial Neural Networks (ANNs) is increased by preferring to reject data, that is not trustful, instead of risking a misclassification. For this purpose a model is proposed that is able to recognize data, which differs significantly from the training data, during inference. The proposed model observes all activations of the hidden layers, as well as input and output layers of an ANN in a grey-box view. To make ANNs more robust in safety critical applications, this model can be used to reject flawed data, that is suspected to decrease the accuracy of the model. If this information is logged during inference, it can be used to improve the model, by training it specifically with the missing information. An experiment on the MNIST dataset is conducted and its results are discussed.

Export metadata

Additional Services

Share in Twitter Search Google Scholar Statistics
Metadaten
Author:Andreas Gschossmann, Simon JobstORCiD, Jürgen MottokORCiDGND, Rudolf Bierl
URL / DOI:https://ieeexplore.ieee.org/document/8836211
Parent Title (English):ARCS Workshop 2019; 32nd International Conference on Architecture of Computing Systems, 20-21 May 2019, Copenhagen, Denmark
Publisher:VDE-Verlag
Place of publication:Berlin
Document Type:conference proceeding (article)
Language:English
Year of first Publication:2019
Release Date:2021/02/19
First Page:153
Last Page:157
Institutes:Fakultät Angewandte Natur- und Kulturwissenschaften
Fakultät Elektro- und Informationstechnik
Fakultät Angewandte Natur- und Kulturwissenschaften / Sensorik-Applikationszentrum (SappZ)
Fakultät Elektro- und Informationstechnik / Laboratory for Safe and Secure Systems (LAS3)
Begutachtungsstatus:peer-reviewed
research focus:Digitalisierung