Biomedical Image Classification via Dynamically Early Stopped Artificial Neural Network
It is well known that biomedical imaging analysis plays a crucial role in the healthcare sector and produces a huge quantity of data. These data can be exploited to study diseases and their evolution in a deeper way or to predict their onsets. In particular, image classification represents one of th...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-10-01
|
| Series: | Algorithms |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1999-4893/15/10/386 |
| _version_ | 1797476027478310912 |
|---|---|
| author | Giorgia Franchini Micaela Verucchi Ambra Catozzi Federica Porta Marco Prato |
| author_facet | Giorgia Franchini Micaela Verucchi Ambra Catozzi Federica Porta Marco Prato |
| author_sort | Giorgia Franchini |
| collection | DOAJ |
| description | It is well known that biomedical imaging analysis plays a crucial role in the healthcare sector and produces a huge quantity of data. These data can be exploited to study diseases and their evolution in a deeper way or to predict their onsets. In particular, image classification represents one of the main problems in the biomedical imaging context. Due to the data complexity, biomedical image classification can be carried out by trainable mathematical models, such as artificial neural networks. When employing a neural network, one of the main challenges is to determine the optimal duration of the training phase to achieve the best performance. This paper introduces a new adaptive early stopping technique to set the optimal training time based on dynamic selection strategies to fix the learning rate and the mini-batch size of the stochastic gradient method exploited as the optimizer. The numerical experiments, carried out on different artificial neural networks for image classification, show that the developed adaptive early stopping procedure leads to the same literature performance while finalizing the training in fewer epochs. The numerical examples have been performed on the CIFAR100 dataset and on two distinct MedMNIST2D datasets which are the large-scale lightweight benchmark for biomedical image classification. |
| first_indexed | 2024-03-09T20:53:05Z |
| format | Article |
| id | doaj.art-f25687b4d1a94100a8c321c4dc2a59f1 |
| institution | Directory Open Access Journal |
| issn | 1999-4893 |
| language | English |
| last_indexed | 2024-03-09T20:53:05Z |
| publishDate | 2022-10-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Algorithms |
| spelling | doaj.art-f25687b4d1a94100a8c321c4dc2a59f12023-11-23T22:30:43ZengMDPI AGAlgorithms1999-48932022-10-01151038610.3390/a15100386Biomedical Image Classification via Dynamically Early Stopped Artificial Neural NetworkGiorgia Franchini0Micaela Verucchi1Ambra Catozzi2Federica Porta3Marco Prato4Department of Physics, Informatics and Mathematics, University of Modena and Reggio Emilia, 41125 Modena, ItalyDepartment of Physics, Informatics and Mathematics, University of Modena and Reggio Emilia, 41125 Modena, ItalyDepartment of Physics, Informatics and Mathematics, University of Modena and Reggio Emilia, 41125 Modena, ItalyDepartment of Physics, Informatics and Mathematics, University of Modena and Reggio Emilia, 41125 Modena, ItalyDepartment of Physics, Informatics and Mathematics, University of Modena and Reggio Emilia, 41125 Modena, ItalyIt is well known that biomedical imaging analysis plays a crucial role in the healthcare sector and produces a huge quantity of data. These data can be exploited to study diseases and their evolution in a deeper way or to predict their onsets. In particular, image classification represents one of the main problems in the biomedical imaging context. Due to the data complexity, biomedical image classification can be carried out by trainable mathematical models, such as artificial neural networks. When employing a neural network, one of the main challenges is to determine the optimal duration of the training phase to achieve the best performance. This paper introduces a new adaptive early stopping technique to set the optimal training time based on dynamic selection strategies to fix the learning rate and the mini-batch size of the stochastic gradient method exploited as the optimizer. The numerical experiments, carried out on different artificial neural networks for image classification, show that the developed adaptive early stopping procedure leads to the same literature performance while finalizing the training in fewer epochs. The numerical examples have been performed on the CIFAR100 dataset and on two distinct MedMNIST2D datasets which are the large-scale lightweight benchmark for biomedical image classification.https://www.mdpi.com/1999-4893/15/10/386image classificationbiomedical imagingearly stoppingartificial neural networkGreenAIhealth care |
| spellingShingle | Giorgia Franchini Micaela Verucchi Ambra Catozzi Federica Porta Marco Prato Biomedical Image Classification via Dynamically Early Stopped Artificial Neural Network Algorithms image classification biomedical imaging early stopping artificial neural network GreenAI health care |
| title | Biomedical Image Classification via Dynamically Early Stopped Artificial Neural Network |
| title_full | Biomedical Image Classification via Dynamically Early Stopped Artificial Neural Network |
| title_fullStr | Biomedical Image Classification via Dynamically Early Stopped Artificial Neural Network |
| title_full_unstemmed | Biomedical Image Classification via Dynamically Early Stopped Artificial Neural Network |
| title_short | Biomedical Image Classification via Dynamically Early Stopped Artificial Neural Network |
| title_sort | biomedical image classification via dynamically early stopped artificial neural network |
| topic | image classification biomedical imaging early stopping artificial neural network GreenAI health care |
| url | https://www.mdpi.com/1999-4893/15/10/386 |
| work_keys_str_mv | AT giorgiafranchini biomedicalimageclassificationviadynamicallyearlystoppedartificialneuralnetwork AT micaelaverucchi biomedicalimageclassificationviadynamicallyearlystoppedartificialneuralnetwork AT ambracatozzi biomedicalimageclassificationviadynamicallyearlystoppedartificialneuralnetwork AT federicaporta biomedicalimageclassificationviadynamicallyearlystoppedartificialneuralnetwork AT marcoprato biomedicalimageclassificationviadynamicallyearlystoppedartificialneuralnetwork |