arXiv:2609.24381
Versatile Quantum Machine Learning with an Ultra-low Power Photonic Quantum Reservoir Computer
Wei Wang, Zan Tang, Menglong Fang, Daiqin Su, Mile Gu, Jayne Thompson, Lip Ket Chin, Hong Cai, Leong-Chuan Kwek, Ai-Qun Liu
Abstract
Integrated photonic microprocessors provide high-bandwidth, massively parallel linear computation, but realizing nonlinear feature maps and temporal memory remain key challenges for machine learning. Conventional approaches rely on active tuning and additional nonlinear elements, increasing architectural complexity and power overhead. Here we demonstrate an integrated photonic quantum reservoir computer that achieves nonlinear mapping, fading memory, and task versatility without active tuning of the reservoir core. The same chip supports accurate static classification, dynamic prediction, and stable autonomous forecasting, establishing broad utility across both classification and temporal inference tasks. Competitive performance is retained in the zero-bias state, where all on-chip phase shifters are unpowered, eliminating active control and reducing computational power consumption to zero. This passive operation highlights a scalable route to multifunctional machine-learning hardware, where large-scale photonic quantum processors can be repurposed as reservoirs without reconfiguring their internal optical networks. By combining quantum-state encoding, multimode interferometric mixing, and photon-statistical readout, this architecture provides a physically grounded paradigm for low-power, large-scale quantum reservoir computing.
AI Summary
The paper addresses how integrated photonic hardware can provide nonlinear feature mapping and temporal memory for machine learning without actively tuning the optical network or adding nonlinear components. It reports an experimental photonic quantum reservoir computer that combines quantum-state encoding, multimode interferometric mixing, and photon-statistical readout, using a fixed reservoir core for both static and temporal tasks. The abstract reports accurate static classification, dynamic prediction, and stable autonomous forecasting on the same chip. It also states that competitive performance persists in a zero-bias configuration, with all on-chip phase shifters unpowered. This establishes the central reported hardware result: the reservoir can operate without electrical power for its phase-shifter control. The abstract describes this as reducing computational power consumption to zero, but does not provide a system-level energy accounting covering optical sources, state preparation, detection, or readout processing. Only the title and abstract are available. No numerical accuracies, prediction errors, forecasting horizons, chip dimensions, photon rates, or comparisons with classical reservoirs are supplied, so the magnitude of the reported performance and its experimental basis cannot be assessed here. The abstract does not specify how fading memory is implemented or quantified, nor whether any computational advantage depends on quantum resources. It states no explicit limitations and provides insufficient detail to evaluate scalability or total energy efficiency.
Written by gpt-6.1-sol from the abstract only. Automated: it can be wrong.