Quantum Neural Networks
A Path to Lower Emissions Through Fuel Consumption Prediction in Shipping
S. F. Chien (Axon Logic, MIMOS Berhad)
Julien J.M. Hermans (TU Delft - Ship Design, Production and Operations)
Austin A. Kana (TU Delft - Ship Design, Production and Operations)
Charilaos C. Zarakovitis (Axon Logic)
Stathis Zavvos (VLTN BV)
H. S. Lim (Multimedia University, Axon Logic)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
This paper proposes Quantum Neural Networks (QNNs) as a data-driven approach for predicting fuel consumption. We utilize various layer architecture designs available in the Torchquantum framework, including both entangled and non-entangled circuit designs. In general, QNNs can achieve comparable Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE) with significantly fewer trainable parameters. Neither pure QNNs nor hybrid QNN models exhibit the underfitting tendencies seen in classical neural networks (CNNs). Notably, one of the most significant findings of this work is that hybridizing or”dressing” the quantum circuit leads to substantial improvements in RMSE and MAPE for pure QNNs. These promising results suggest potential optimizations for reducing emissions in green shipping.