Privacy-Preserving Collaborative AI for Distributed Deep Learning with Cross-sectional Data
- Resource Type
- Conference
- Authors
- Iqbal, Saeed; Qureshi, Adnan N.; Aurangzeb, Khursheed; Javeed, Khalid
- Source
- 2023 5th International Conference on Bio-engineering for Smart Technologies (BioSMART) Bio-engineering for Smart Technologies (BioSMART), 2023 5th International Conference on. :1-4 Jun, 2023
- Subject
- Bioengineering
Computing and Processing
General Topics for Engineers
Robotics and Control Systems
Signal Processing and Analysis
Deep learning
Data privacy
Image analysis
Federated learning
Medical services
Skin
Convolutional neural networks
Privacy Preserving
Collaborative AI
Cross-sectional Data
- Language
- ISSN
- 2831-4352
The article discusses the challenges of using deep learning in healthcare due to the lack of extensive medical datasets and concerns about confidentiality and privacy. The article then focuses on the analysis of skin lesion images, which are complex and difficult to classify. The article presents a unique CNN model (skin-net) using a progressively private federated learning architecture to potentially address these issues. Recent advancements in medical image analysis using convolutional neural networks and federated learning techniques have shown promising results (96.19 accuracy, 96.37 sensitivity and 96.53 specificity) in accurately classifying skin diseases while maintaining data privacy. Using these methods offers a potentially effective way to examine human skin while improving the model’s generalization performance through image augmentation.