2418

View Abstract / View Presentation

Federated Multi-task Image Classification on Heterogeneous Medical data with Privacy Perversing

Shenjun Zhong¹, Adam Morris², Zhaolin Chen¹, and Gary Egan¹
¹Monash Biomedical Imaging, Monash University, Australia, Melbourne, Australia, ²Monash eResearch Center, Monash University, Australia, Melbourne, Australia

This work applied the multi-task learning process in federated learning settings, and validated the performance of the co-trained models that could be used for downstream medical image analysis tasks.

Figure 1. Federate Learning Workflow

Figure 3. ACC Comparison of Federated Trained Models with Baseline Models