Reservoir Computing Techniques Using Tensor Networks

Abstract

Reservoir computing (RC), traditionally based on echo state networks and liquid state machines, shows great potential in modeling dynamic time-series data like weather and astronomical predictions. However, these frameworks are not suitable for quantum dynamics-based RC. Tensor networks (TNs) are well suited for modeling quantum dynamics because they can efficiently model quantum information and entanglement. In this work, we propose a novel randomized TN-based RC scheme, demonstrating its validity through various case studies. Our results show superior performance compared to traditional ESN models, laying the groundwork for further exploration of quantum reservoir computing.