Continuous Variable Single Mode Quantum Decoder for Image Reconstruction and Denoising
AI Breakdown
Get a structured breakdown of this paper — what it's about, the core idea, and key takeaways for the field.
Abstract
Quantum computation using optical modes has been well-established in its ability to construct deep neural networks. We introduce a model that is the quantum analogue of the classical autoencoder - a neural network model that can reconstruct its input via dimensionality reduction and expansion through the phase-space formulation of quantum mechanics. The hallmark of the continuous-variable (CV) model is its ability to forge non-linear functions using a set of gates that allows it to remain completely unitary. We leverage this property of the CV model to encode and decode - classical information and demonstrate denoising applications using parallel single mode photonic circuits. The proposed model exemplifies that the appropriate photonic hardware can be integrated with present day optical communication systems to meet our information processing requirements. Here, using the Strawberry Fields software library on the MNIST dataset of handwritten digits, we demonstrate the adaptability of the network to learn classical information to fidelities of greater than 99.98\%.