An Experimental Analysis on the Application of Quantum Federated Learning in Healthcare Analytics with Variational Parameter Set-up

The paradigm shift of moving from traditional machine learning models to distributed learning using Federated learning has gained its importance in various real-time applications especially in applications that requires privacy preservation as the primary criterion. However, for complex real-time data, the computational limitations and communication bottlenecks in the current federated learning environment has made the advancements of applying Quantum computing in Federated Learning. Quantum Federated Learning (QFL) is combining FL with Quantum computing. In this paper, QFL is applied for a real-time healthcare analytics dataset to assess the applicability of Quantum computing in such cases. From our implementation, we realized that, even though QFL has several advantages in terms of distributed model training with quantum-based communications, it suffers from accuracy concerns as the current implementation is experimented in the classical system configuration and not in the QPU or Quantum computer set-up. We concluded that, if the Quantum computers or QPU are utilized for such complex real-time dataset, we can get the real advancements QFL.