Paper Title :Bayesian Hyperparameter Optimization for Custom Convolutional Neural Networks in Skin Cancer Recognition
Author :Babak Mostafavi, Mahdi Amirsardari
Article Citation :Babak Mostafavi ,Mahdi Amirsardari ,
(2024 ) " Bayesian Hyperparameter Optimization for Custom Convolutional Neural Networks in Skin Cancer Recognition " ,
International Journal of Electrical, Electronics and Data Communication (IJEEDC) ,
pp. 22-28,
Volume-12,Issue-1
Abstract : Skin cancer is a significant public health concern, and early detection is crucial for effective treatment. Deep
learning-based approaches, particularly Convolutional Neural Networks (CNNs), have shown promise in automating skin
cancer recognition from medical images. However, designing an optimal CNN architecture with the right hyperparameters
remains a challenging and time-consuming task. This article presents an innovative approach to address this challenge using
Bayesian Hyperparameter Optimization. Instead of relying on pre-trained models, we explore the creation of custom CNN
architectures tailored specifically for skin cancer recognition. Bayesian optimization is employed to systematically search
and discover the optimal combination of hyperparameters, significantly enhancing model performance. We discuss the
development of a custom CNN, data preprocessing techniques, and the incorporation of Bayesian optimization to fine-tune
hyperparameters. Our experiments demonstrate the effectiveness of this approach, resulting in a skin cancer recognition
model with superior accuracy compared to conventional methods. Through this work, we provide a valuable contribution to
the field of dermatology by showcasing the potential of custom CNNs and Bayesian optimization in automating skin cancer
diagnosis. Our findings open up new avenues for improving the accuracy and efficiency of skin cancer recognition systems,
with broader implications for the field of medical image analysis.
Keywords - skin cancer, Image processing, convolutional neural networks, Bayesian optimization
Type : Research paper
Published : Volume-12,Issue-1
DOIONLINE NO - IJEEDC-IRAJ-DOIONLINE-20463
View Here
Copyright: © Institute of Research and Journals
|
![](images/permissions-active.png) |
| |
![](images/pdf_icon.png) |
PDF |
| |
Viewed - 17 |
| |
Published on 2024-03-27 |
|