Autoencoder-Based Semantic Novelty Detection: Towards Dependable AI-Based Systems
Many autonomous systems, such as driverless taxis, perform safety-critical functions. Autonomous systems employ artificial intelligence (AI) techniques, specifically for environmental perception. Engineers cannot completely test or formally verify AI-based autonomous systems. The accuracy of AI-base...
Guardado en:
Autores principales: | , , |
---|---|
Formato: | article |
Lenguaje: | EN |
Publicado: |
MDPI AG
2021
|
Materias: | |
Acceso en línea: | https://doaj.org/article/99dbbcd60bfe4bf99211cf25abf9360e |
Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
id |
oai:doaj.org-article:99dbbcd60bfe4bf99211cf25abf9360e |
---|---|
record_format |
dspace |
spelling |
oai:doaj.org-article:99dbbcd60bfe4bf99211cf25abf9360e2021-11-11T14:59:46ZAutoencoder-Based Semantic Novelty Detection: Towards Dependable AI-Based Systems10.3390/app112198812076-3417https://doaj.org/article/99dbbcd60bfe4bf99211cf25abf9360e2021-10-01T00:00:00Zhttps://www.mdpi.com/2076-3417/11/21/9881https://doaj.org/toc/2076-3417Many autonomous systems, such as driverless taxis, perform safety-critical functions. Autonomous systems employ artificial intelligence (AI) techniques, specifically for environmental perception. Engineers cannot completely test or formally verify AI-based autonomous systems. The accuracy of AI-based systems depends on the quality of training data. Thus, novelty detection, that is, identifying data that differ in some respect from the data used for training, becomes a safety measure for system development and operation. In this study, we propose a new architecture for autoencoder-based semantic novelty detection with two innovations: architectural guidelines for a semantic autoencoder topology and a semantic error calculation as novelty criteria. We demonstrate that such a semantic novelty detection outperforms autoencoder-based novelty detection approaches known from the literature by minimizing false negatives.Andreas RauschAzarmidokht Motamedi SedehMeng ZhangMDPI AGarticlesafety engineeringautonomous systemperceptionartificial intelligenceautoencodernovelty detectionTechnologyTEngineering (General). Civil engineering (General)TA1-2040Biology (General)QH301-705.5PhysicsQC1-999ChemistryQD1-999ENApplied Sciences, Vol 11, Iss 9881, p 9881 (2021) |
institution |
DOAJ |
collection |
DOAJ |
language |
EN |
topic |
safety engineering autonomous system perception artificial intelligence autoencoder novelty detection Technology T Engineering (General). Civil engineering (General) TA1-2040 Biology (General) QH301-705.5 Physics QC1-999 Chemistry QD1-999 |
spellingShingle |
safety engineering autonomous system perception artificial intelligence autoencoder novelty detection Technology T Engineering (General). Civil engineering (General) TA1-2040 Biology (General) QH301-705.5 Physics QC1-999 Chemistry QD1-999 Andreas Rausch Azarmidokht Motamedi Sedeh Meng Zhang Autoencoder-Based Semantic Novelty Detection: Towards Dependable AI-Based Systems |
description |
Many autonomous systems, such as driverless taxis, perform safety-critical functions. Autonomous systems employ artificial intelligence (AI) techniques, specifically for environmental perception. Engineers cannot completely test or formally verify AI-based autonomous systems. The accuracy of AI-based systems depends on the quality of training data. Thus, novelty detection, that is, identifying data that differ in some respect from the data used for training, becomes a safety measure for system development and operation. In this study, we propose a new architecture for autoencoder-based semantic novelty detection with two innovations: architectural guidelines for a semantic autoencoder topology and a semantic error calculation as novelty criteria. We demonstrate that such a semantic novelty detection outperforms autoencoder-based novelty detection approaches known from the literature by minimizing false negatives. |
format |
article |
author |
Andreas Rausch Azarmidokht Motamedi Sedeh Meng Zhang |
author_facet |
Andreas Rausch Azarmidokht Motamedi Sedeh Meng Zhang |
author_sort |
Andreas Rausch |
title |
Autoencoder-Based Semantic Novelty Detection: Towards Dependable AI-Based Systems |
title_short |
Autoencoder-Based Semantic Novelty Detection: Towards Dependable AI-Based Systems |
title_full |
Autoencoder-Based Semantic Novelty Detection: Towards Dependable AI-Based Systems |
title_fullStr |
Autoencoder-Based Semantic Novelty Detection: Towards Dependable AI-Based Systems |
title_full_unstemmed |
Autoencoder-Based Semantic Novelty Detection: Towards Dependable AI-Based Systems |
title_sort |
autoencoder-based semantic novelty detection: towards dependable ai-based systems |
publisher |
MDPI AG |
publishDate |
2021 |
url |
https://doaj.org/article/99dbbcd60bfe4bf99211cf25abf9360e |
work_keys_str_mv |
AT andreasrausch autoencoderbasedsemanticnoveltydetectiontowardsdependableaibasedsystems AT azarmidokhtmotamedisedeh autoencoderbasedsemanticnoveltydetectiontowardsdependableaibasedsystems AT mengzhang autoencoderbasedsemanticnoveltydetectiontowardsdependableaibasedsystems |
_version_ |
1718437913399656448 |