Federated Learning on Multiclass Classification June 2022

Implemented a federated learning algorithm as a security and privacy-preserving approach to training machine learning models.

Technical stack Used in the Project -

• The CIFAR dataset served as the training data, involving 20 client nodes, each engaging in training activities based on the VGG-19 model.
• The central global model collected weight updates from six randomly selected client models, averaging the contributions, and disseminating the updated global model to all participating clients.
• Upon the successful completion of the training process, achieved a commendable accuracy rate of 78%, showcasing the effectiveness of this federated learning approach in preserving data security and privacy while maintaining model performance.

The Github code is here