Cargando…
Sensitivity of neural networks to corruption of image classification
Artificial intelligence (AI) systems are extensively used today in many fields. In the field of medicine, AI-systems are especially used for the segmentation and classification of medical images. As reliance on such AI-systems increases, it is important to verify that these systems are dependable an...
Autores principales: | , , |
---|---|
Formato: | Online Artículo Texto |
Lenguaje: | English |
Publicado: |
Springer International Publishing
2021
|
Materias: | |
Acceso en línea: | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7985580/ https://www.ncbi.nlm.nih.gov/pubmed/34790945 http://dx.doi.org/10.1007/s43681-021-00049-0 |
_version_ | 1783668278592274432 |
---|---|
author | Kaplan, Shimon Handelman, Doron Handelman, Amir |
author_facet | Kaplan, Shimon Handelman, Doron Handelman, Amir |
author_sort | Kaplan, Shimon |
collection | PubMed |
description | Artificial intelligence (AI) systems are extensively used today in many fields. In the field of medicine, AI-systems are especially used for the segmentation and classification of medical images. As reliance on such AI-systems increases, it is important to verify that these systems are dependable and not sensitive to bias or other types of errors that may severely affect users and patients. This work investigates the sensitivity of the performance of AI-systems to labeling errors. Such investigation is performed by simulating intentional mislabeling of training images according to different values of a new parameter called “mislabeling balance” and a “corruption” parameter, and then measuring the accuracy of the AI-systems for every value of these parameters. The issues investigated in this work include the amount (percentage) of errors from which a substantial adverse effect on the performance of the AI-systems can be observed, and how unreliable labeling can be done in the training stage. The goals of this work are to raise ethical concerns regarding the various types of errors that can possibly find their way into AI-systems, to demonstrate the effect of training errors, and to encourage development of techniques that can cope with the problem of errors, especially for AI-systems that perform sensitive medical-related tasks. |
format | Online Article Text |
id | pubmed-7985580 |
institution | National Center for Biotechnology Information |
language | English |
publishDate | 2021 |
publisher | Springer International Publishing |
record_format | MEDLINE/PubMed |
spelling | pubmed-79855802021-03-23 Sensitivity of neural networks to corruption of image classification Kaplan, Shimon Handelman, Doron Handelman, Amir AI Ethics Original Research Artificial intelligence (AI) systems are extensively used today in many fields. In the field of medicine, AI-systems are especially used for the segmentation and classification of medical images. As reliance on such AI-systems increases, it is important to verify that these systems are dependable and not sensitive to bias or other types of errors that may severely affect users and patients. This work investigates the sensitivity of the performance of AI-systems to labeling errors. Such investigation is performed by simulating intentional mislabeling of training images according to different values of a new parameter called “mislabeling balance” and a “corruption” parameter, and then measuring the accuracy of the AI-systems for every value of these parameters. The issues investigated in this work include the amount (percentage) of errors from which a substantial adverse effect on the performance of the AI-systems can be observed, and how unreliable labeling can be done in the training stage. The goals of this work are to raise ethical concerns regarding the various types of errors that can possibly find their way into AI-systems, to demonstrate the effect of training errors, and to encourage development of techniques that can cope with the problem of errors, especially for AI-systems that perform sensitive medical-related tasks. Springer International Publishing 2021-03-23 2021 /pmc/articles/PMC7985580/ /pubmed/34790945 http://dx.doi.org/10.1007/s43681-021-00049-0 Text en © The Author(s), under exclusive licence to Springer Nature Switzerland AG 2021 This article is made available via the PMC Open Access Subset for unrestricted research re-use and secondary analysis in any form or by any means with acknowledgement of the original source. These permissions are granted for the duration of the World Health Organization (WHO) declaration of COVID-19 as a global pandemic. |
spellingShingle | Original Research Kaplan, Shimon Handelman, Doron Handelman, Amir Sensitivity of neural networks to corruption of image classification |
title | Sensitivity of neural networks to corruption of image classification |
title_full | Sensitivity of neural networks to corruption of image classification |
title_fullStr | Sensitivity of neural networks to corruption of image classification |
title_full_unstemmed | Sensitivity of neural networks to corruption of image classification |
title_short | Sensitivity of neural networks to corruption of image classification |
title_sort | sensitivity of neural networks to corruption of image classification |
topic | Original Research |
url | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7985580/ https://www.ncbi.nlm.nih.gov/pubmed/34790945 http://dx.doi.org/10.1007/s43681-021-00049-0 |
work_keys_str_mv | AT kaplanshimon sensitivityofneuralnetworkstocorruptionofimageclassification AT handelmandoron sensitivityofneuralnetworkstocorruptionofimageclassification AT handelmanamir sensitivityofneuralnetworkstocorruptionofimageclassification |