Saptadip Saha
National Institute of Technology
Agartala, India
With
Felix Ringer
Jefferson Lab, Old Dominion University, USA
As part of
REYES 2024 (UC Berkeley)
Continuous Variable Quantum Neural Networks (CV-QNNs) represent a promising frontier in quantum machine learning, offering unique advantages for tasks which require continuous real-valued inputs and outputs. We begin by introducing the fundamental concepts of quantum computing, then transition into CV quantum computing, focusing on qumodes as information carriers, which are analogous to qubits in discrete variable models. We explore both Gaussian and non-Gaussian transformations and the representation of qumodes states in Hilbert space and phase space. By leveraging these properties, we illustrate how to design and implement variational circuits that perform in a manner similar to classical Multilayer Perceptrons (MLP). As a practical application, we apply a CV-QNN to a supervised regression task, specifically focusing on approximating a noisy continuous function. This application showcases a potential use case of CV quantum computing to tackle machine learning problems which require continuous real-valued inputs and outputs. Paper