Feedback-enhanced quantum reservoir computing with weak measurements

Abstract: Quantum reservoir computing (QRC) leverages the natural dynamics of quantum systems to process time-series data efficiently, offering a promising approach for near-term quantum devices. Unlike classical reservoir computing, the efficacy of feedback in QRC has not yet been thoroughly explored. Here, we develop a feedback-enhanced QRC framework with weak measurements. Weak measurements preserve information stored in quantum coherence, while feedback enhances nonlinearity and memory capacity. The implementation of our framework assumes an ensemble quantum system, such as nuclear magnetic resonance. Through linear memory and nonlinear forecasting tasks, we show that our model outperforms conventional QRC approaches in many cases. Our proposed protocol achieves superior performance in systems with small measurement errors and low environmental noise. Furthermore, we theoretically demonstrate that feedback of measurement results reinforces the nonlinearity of the reservoir. These findings highlight the potential of feedback-enhanced QRC for next-generation quantum machine learning applications.