Have you ever imagined how your mobile keyboard predicts the next keyword strokes?
The working/prediction of words is given by an ML algorithm deployed on your mobile and is trained on your local data. As it’s training on your data, you might have a question now, isn’t my privacy lost?
The answer is no. These tech companies use Federated Learning to solve the privacy issues as no data is sent to the primary model. Instead, the ML model will be deployed on your device and trained on the data available, and the model parameters are returned instead of the data. Let’s dive into the working of it, and in this article, we propose a new methodology of clustering similar devices to increase the model’s performance.
List of contents:
- Introduction
- Federated Learning
- Our Framework
- Clustering
- Phases of Training
- Results & Comparision
- Conclusion
#tensorflow #ai #deep-learning #machine-learning #federated-learning
