I had the pleasure of working with Mayank on a recent project, and I am thoroughly impressed with his professionalism and expertise. From the outset, Mayank demonstrated a deep understanding of the project requirements and consistently delivered high-quality work. His attention to detail, commitment to deadlines, and proactive communication made the entire process smooth and efficient.
BAFL: Federated Learning with Base Ablation for Cost Effective Communication
ICPR | IEEE Computer Society Digital Library
Two major challenges faced in this Federated Learning are data heterogeneity and high communication cost. We target the latter and propose a simple approach, BAFL to reduce communication costs. In contrast to the common practice of employing model compression techniques to reduce the total communication cost, we propose a fine-tuning approach to leverage the feature extraction ability of layers at different depths of deep neural networks.
An improved privacy preservation technique in health-cloud
ICT Express, Elsevier
I have proposed a hybrid technique which includes two different inference control techniques, query set size restriction and k-anonymity to ensure individuals’ privacy. A query set size restriction is used to prevent the sensitive data from inference attacks, whereas k-anonymity is implemented to protect the data from linking attacks.
https://doi.org/10.1016/j.icte.2018.10.002