Variational quantum circuits, previously hoped to outperform their classical counterparts, now appear less certain in the realm of time-series forecasting. While fully quantum architectures performed demonstrably worse, the hybrid quantum-classical model proved statistically indistinguishable from its strongest classical analogue on both Gaussian-process and NARMA-10 datasets. This finding underscores the need for rigorous, symmetric evaluation when claiming quantum advantage in machine learning.
Researchers Find No Quantum Advantage in Time-Series Forecasting Models
Variational quantum circuits, previously hoped to outperform their classical counterparts, now appear less certain in the realm of time-series forecasting. While fully quantum architectures performed demonstrably worse, the hybrid quantum-classical model proved statistically indistinguishable from its strongest classical analogue on both Gaussian-process and NARMA-10 datasets. This finding underscores the need for rigorous, symmetric evaluation when claiming quantum advantage in machine learn...