The effect of batch size on the generalizability of the convolutional neural networks on a histopathology dataset
Many hyperparameters have to be tuned to have a robust convolutional neural network that will be able to accurately classify images. One of the most important hyperparameters is the batch size, which is the number of images used to train a single forward and backward pass. In this study, the effect...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2020-12-01
|
| Series: | ICT Express |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2405959519303455 |
| _version_ | 1819092033988460544 |
|---|---|
| author | Ibrahem Kandel Mauro Castelli |
| author_facet | Ibrahem Kandel Mauro Castelli |
| author_sort | Ibrahem Kandel |
| collection | DOAJ |
| description | Many hyperparameters have to be tuned to have a robust convolutional neural network that will be able to accurately classify images. One of the most important hyperparameters is the batch size, which is the number of images used to train a single forward and backward pass. In this study, the effect of batch size on the performance of convolutional neural networks and the impact of learning rates will be studied for image classification, specifically for medical images. To train the network faster, a VGG16 network with ImageNet weights was used in this experiment. Our results concluded that a higher batch size does not usually achieve high accuracy, and the learning rate and the optimizer used will have a significant impact as well. Lowering the learning rate and decreasing the batch size will allow the network to train better, especially in the case of fine-tuning. |
| first_indexed | 2024-12-21T22:49:11Z |
| format | Article |
| id | doaj.art-63e2e82fd5a54df790389e1b73a23ece |
| institution | Directory Open Access Journal |
| issn | 2405-9595 |
| language | English |
| last_indexed | 2024-12-21T22:49:11Z |
| publishDate | 2020-12-01 |
| publisher | Elsevier |
| record_format | Article |
| series | ICT Express |
| spelling | doaj.art-63e2e82fd5a54df790389e1b73a23ece2022-12-21T18:47:36ZengElsevierICT Express2405-95952020-12-0164312315The effect of batch size on the generalizability of the convolutional neural networks on a histopathology datasetIbrahem Kandel0Mauro Castelli1Corresponding author.; Nova Information Management School (NOVA IMS), Universidade Nova de Lisboa, Campus de Campolide, 1070-312, Lisbon, PortugalNova Information Management School (NOVA IMS), Universidade Nova de Lisboa, Campus de Campolide, 1070-312, Lisbon, PortugalMany hyperparameters have to be tuned to have a robust convolutional neural network that will be able to accurately classify images. One of the most important hyperparameters is the batch size, which is the number of images used to train a single forward and backward pass. In this study, the effect of batch size on the performance of convolutional neural networks and the impact of learning rates will be studied for image classification, specifically for medical images. To train the network faster, a VGG16 network with ImageNet weights was used in this experiment. Our results concluded that a higher batch size does not usually achieve high accuracy, and the learning rate and the optimizer used will have a significant impact as well. Lowering the learning rate and decreasing the batch size will allow the network to train better, especially in the case of fine-tuning.http://www.sciencedirect.com/science/article/pii/S2405959519303455Convolutional neural networksDeep learningImage classificationMedical imagesBatch size |
| spellingShingle | Ibrahem Kandel Mauro Castelli The effect of batch size on the generalizability of the convolutional neural networks on a histopathology dataset ICT Express Convolutional neural networks Deep learning Image classification Medical images Batch size |
| title | The effect of batch size on the generalizability of the convolutional neural networks on a histopathology dataset |
| title_full | The effect of batch size on the generalizability of the convolutional neural networks on a histopathology dataset |
| title_fullStr | The effect of batch size on the generalizability of the convolutional neural networks on a histopathology dataset |
| title_full_unstemmed | The effect of batch size on the generalizability of the convolutional neural networks on a histopathology dataset |
| title_short | The effect of batch size on the generalizability of the convolutional neural networks on a histopathology dataset |
| title_sort | effect of batch size on the generalizability of the convolutional neural networks on a histopathology dataset |
| topic | Convolutional neural networks Deep learning Image classification Medical images Batch size |
| url | http://www.sciencedirect.com/science/article/pii/S2405959519303455 |
| work_keys_str_mv | AT ibrahemkandel theeffectofbatchsizeonthegeneralizabilityoftheconvolutionalneuralnetworksonahistopathologydataset AT maurocastelli theeffectofbatchsizeonthegeneralizabilityoftheconvolutionalneuralnetworksonahistopathologydataset AT ibrahemkandel effectofbatchsizeonthegeneralizabilityoftheconvolutionalneuralnetworksonahistopathologydataset AT maurocastelli effectofbatchsizeonthegeneralizabilityoftheconvolutionalneuralnetworksonahistopathologydataset |