Publications ›› Papers ›› Data Privacy

Optimizing Federated Learning using Remote Embeddings for Graph Neural Networks

Authors: Pranjal Naman and Yogesh Simmhan, Know more Graph Neural Networks (GNNs) have experienced rapid advancements in recent years due to their ability to learn meaningful representations from graph data structures. Federated Learning (FL) has emerged as a viable machine learning approach for training a shared model on decentralized data, addressing privacy concerns while leveraging parallelism. Existing methods that address […]

Minimizing Layerwise Activation Norm Improves Generalization in Federated Learning

Authors: M. Yashwanth, G. K. Nayak, H. Rangwani, A. Singh, R. V. Babu, A. Chakraborty Know more Federated Learning (FL) is an emerging machine learning framework that enables multiple clients (coordinated by a server) to collaboratively train a global model by aggregating the locally trained models without sharing any client’s training data. It has been observed […]

Continual Mean Estimation Under User-Level Privacy

Authors: A. J. George, L. Ramesh, A. V. Singh and H. Tyagi Know more We consider the problem of continually releasing an estimate of the population mean of a stream of samples that is user-level differentially private (DP). At each time instant, a user contributes a sample, and the users can arrive in arbitrary order. […]

User-Level Differentially Private Mean Estimation for Real-World Datasets

Authors: V. A. Rameshwar, A. Tandon, and A. Sharma Know more In this work, we provide rigorous theoretical justifications for the performance trends of well-known clipping-based algorithms on real-world ITMS and i.i.d. synthetic datasets. An important contribution of this work is the formalization and explicit computation of the “worst-case estimation error” incurred by a canonical […]

Mean Estimation with User-Level Privacy for Spatio-Temporal IoT Datasets

Authors: P. Gupta, V. A. Rameshwar, A. Tandon and N. Chakraborty Know more This paper considers the problem of the private release of sample means of speed values from traffic datasets. Our key contribution is the development of user-level differentially private algorithms that incorporate carefully chosen parameter values to ensure low estimation errors on real-world […]