Dinkar Dhawale

Title: Dinkar Dhawale: Inventor Advancing Privacy in Federated Learning.

Introduction

Dinkar Dhawale is an inventor based in Bengaluru, India, whose patent work addresses privacy protection in federated learning. He is associated with Intel Corporation and has 1 patent listed in the information provided.

Latest Patents

His patent, “Mitigating Private Data Leakage in a Federated Learning System,” describes a method for protecting data privacy in a federated learning environment. The process receives a model trained by a federated server using public data and global model parameters, then trains it with local shareable data to generate local model parameters. Those parameters are obscured before being sent to the server. The method also describes modifying the model to classify private data and training it with that data using the local parameters, which are then updated.

Career Highlights

Dhawale’s described invention focuses on balancing collaborative model training with safeguards for sensitive information. By distinguishing between shareable and private data and obscuring model parameters, the approach aims to reduce the risk of private data leakage during federated learning.

Collaborations

The coworkers identified in the provided information are Jerome Anand and Anil Kumar Bangalore Rajashekar. The information does not specify their roles in the patent or their relationship to Dhawale’s work.

Conclusion

Dinkar Dhawale’s patent work highlights a practical approach to privacy protection in federated learning, with a focus on limiting exposure of sensitive data while enabling model development.