Minimalistic and Scalable Quantum Reservoir Computing Enhanced with Feedback

Summary

Quantum Reservoir Computing (QRC) uses quantum systems to perform complex computational tasks with exceptional efficiency and reduced energy consumption. A minimalistic QRC framework utilizing a few two-level atoms in a single-mode optical cavity, combined with continuous quantum measurements, achieves high computational expressivity with minimal hardware.

Highlights

• QRC leverages quantum systems for complex computations with efficiency and low energy consumption.
• A minimalistic QRC framework uses a few two-level atoms in an optical cavity with continuous measurements.
• The framework includes reservoir feedback and polynomial regression for high expressivity.
• The performance of QRC is evaluated through tasks like predicting chaotic time-series data and classifying sine-square waveforms.
• Increasing the number of atoms and incorporating feedback mechanisms and polynomial regression significantly enhances QRC performance.
• QRC outperforms classical reservoir computing (CRC) due to its exponentially increased number of quantum basis states.
• The scalability of QRC is convenient compared to reservoirs built upon quantum networks.

Key Insights

• QRC's minimalistic framework achieves high computational expressivity with minimal hardware by incorporating feedback mechanisms and polynomial regression, making it an efficient approach for complex computations.
• The exponential scaling of quantum basis states in QRC underlies its advantage over classical reservoir computing, enabling it to capture more complex features in the input data.
• The scalability of QRC is more convenient compared to reservoirs built upon quantum networks, as adding more atoms into the cavity naturally couples them with the cavity field and the rest of the atoms.
• QRC's performance enhancement due to increased atoms and feedback mechanisms tends to saturate, indicating that the maximum performance achievable with the current sample size has been attained.
• The choice of feedback channels has a smaller effect on performance than the number of atoms, confirming the generality of the result.
• The comparison between QRC and CRC highlights the inherent connections and collaborations between the measured atom and the added atoms in QRC, leading to its superior performance.
• The effect of solely increasing the dimension of the QRC Hilbert space demonstrates the advantage of QRC over CRC, even when the number of measured readouts is fixed.

Mindmap

Citation

Zhu, C., Ehlers, P. J., Nurdin, H. I., & Soh, D. (2024). Minimalistic and Scalable Quantum Reservoir Computing Enhanced with Feedback (Version 1). arXiv. https://doi.org/10.48550/ARXIV.2412.17817