Evolving Hybrid Quantum-Classical GRU Architectures for Multivariate Time Series
Published in 2024 IEEE 34th International Workshop on Machine Learning for Signal Processing (MLSP), 2024
Recommended citation: Francesca De Falco, Leonardo Lavagna, Andrea Ceschini, Antonello Rosato, Massimo Panella: Evolving Hybrid Quantum-Classical GRU Architectures for Multivariate Time Series, in the 2024 IEEE 34th International Workshop on Machine Learning for Signal Processing (MLSP). https://inspirehep.net/literature/2849278
We study an hybrid quantum-classical Gated Recurrent Unit (GRU) used to infer information about multidimensional time series. The main idea is to modify the GRU dynamics by incorporating a quantum reset, update, and output step in the classical GRU cell with a VQC. The results are incouraging, but are obtained on simulations, thus further study will be necessary.
Blog post available here.