Identifying Bias in Deep Neural Networks Using Image Transforms
CNNs have become one of the most commonly used computational tools in the past two decades. One of the primary downsides of CNNs is that they work as a “black box”, where the user cannot necessarily know how the image data are analyzed, and therefore needs to rely on empirical evaluation to test the...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2024-12-01
|
| Series: | Computers |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2073-431X/13/12/341 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850059036842524672 |
|---|---|
| author | Sai Teja Erukude Akhil Joshi Lior Shamir |
| author_facet | Sai Teja Erukude Akhil Joshi Lior Shamir |
| author_sort | Sai Teja Erukude |
| collection | DOAJ |
| description | CNNs have become one of the most commonly used computational tools in the past two decades. One of the primary downsides of CNNs is that they work as a “black box”, where the user cannot necessarily know how the image data are analyzed, and therefore needs to rely on empirical evaluation to test the efficacy of a trained CNN. This can lead to hidden biases that affect the performance evaluation of neural networks, but are difficult to identify. Here we discuss examples of such hidden biases in common and widely used benchmark datasets, and propose techniques for identifying dataset biases that can affect the standard performance evaluation metrics. One effective approach to identify dataset bias is to perform image classification by using merely blank background parts of the original images. However, in some situations, a blank background in the images is not available, making it more difficult to separate foreground or contextual information from the bias. To overcome this, we propose a method to identify dataset bias without the need to crop background information from the images. The method is based on applying several image transforms to the original images, including Fourier transform, wavelet transforms, median filter, and their combinations. These transforms are applied to recover background bias information that CNNs use to classify images. These transformations affect the contextual visual information in a different manner than it affects the systemic background bias. Therefore, the method can distinguish between contextual information and the bias, and can reveal the presence of background bias even without the need to separate sub-image parts from the blank background of the original images. The code used in the experiments is publicly available. |
| format | Article |
| id | doaj-art-91bc7dc938d348dba23b3fbd736376c1 |
| institution | DOAJ |
| issn | 2073-431X |
| language | English |
| publishDate | 2024-12-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Computers |
| spelling | doaj-art-91bc7dc938d348dba23b3fbd736376c12025-08-20T02:50:59ZengMDPI AGComputers2073-431X2024-12-01131234110.3390/computers13120341Identifying Bias in Deep Neural Networks Using Image TransformsSai Teja Erukude0Akhil Joshi1Lior Shamir2Department of Computer Science, Kansas State University, Manhattan, KS 66502, USADepartment of Computer Science, Kansas State University, Manhattan, KS 66502, USADepartment of Computer Science, Kansas State University, Manhattan, KS 66502, USACNNs have become one of the most commonly used computational tools in the past two decades. One of the primary downsides of CNNs is that they work as a “black box”, where the user cannot necessarily know how the image data are analyzed, and therefore needs to rely on empirical evaluation to test the efficacy of a trained CNN. This can lead to hidden biases that affect the performance evaluation of neural networks, but are difficult to identify. Here we discuss examples of such hidden biases in common and widely used benchmark datasets, and propose techniques for identifying dataset biases that can affect the standard performance evaluation metrics. One effective approach to identify dataset bias is to perform image classification by using merely blank background parts of the original images. However, in some situations, a blank background in the images is not available, making it more difficult to separate foreground or contextual information from the bias. To overcome this, we propose a method to identify dataset bias without the need to crop background information from the images. The method is based on applying several image transforms to the original images, including Fourier transform, wavelet transforms, median filter, and their combinations. These transforms are applied to recover background bias information that CNNs use to classify images. These transformations affect the contextual visual information in a different manner than it affects the systemic background bias. Therefore, the method can distinguish between contextual information and the bias, and can reveal the presence of background bias even without the need to separate sub-image parts from the blank background of the original images. The code used in the experiments is publicly available.https://www.mdpi.com/2073-431X/13/12/341biasconvolutional neural networksmachine learningexperimental design |
| spellingShingle | Sai Teja Erukude Akhil Joshi Lior Shamir Identifying Bias in Deep Neural Networks Using Image Transforms Computers bias convolutional neural networks machine learning experimental design |
| title | Identifying Bias in Deep Neural Networks Using Image Transforms |
| title_full | Identifying Bias in Deep Neural Networks Using Image Transforms |
| title_fullStr | Identifying Bias in Deep Neural Networks Using Image Transforms |
| title_full_unstemmed | Identifying Bias in Deep Neural Networks Using Image Transforms |
| title_short | Identifying Bias in Deep Neural Networks Using Image Transforms |
| title_sort | identifying bias in deep neural networks using image transforms |
| topic | bias convolutional neural networks machine learning experimental design |
| url | https://www.mdpi.com/2073-431X/13/12/341 |
| work_keys_str_mv | AT saitejaerukude identifyingbiasindeepneuralnetworksusingimagetransforms AT akhiljoshi identifyingbiasindeepneuralnetworksusingimagetransforms AT liorshamir identifyingbiasindeepneuralnetworksusingimagetransforms |