Asynchronous photonic time-delay reservoir computing.

Area of Science:

• * Nonlinear dynamics
• * Computational neuroscience
• * Optical computing

Background:

• * Traditional time-delay reservoir computing relies on synchronous clock cycles and feedback loop delays, limiting hardware flexibility.
• * Semiconductor lasers with optical feedback offer a potential platform for novel computing paradigms.

Purpose of the Study:

• * To demonstrate an asynchronous reservoir computing network using a semiconductor laser with optical feedback.
• * To investigate the performance of this asynchronous network under varying operational conditions.
• * To compare its predictive capabilities against synchronous reservoir computing counterparts.

Main Methods:

• * Implementation of an asynchronous reservoir computing network utilizing a semiconductor laser and an optical feedback loop.
• * Experimental investigation of network performance across diverse operating parameters.
• * Evaluation using the Santa Fe chaotic time series prediction task.

Main Results:

• * The asynchronous reservoir computing network operated with a clock cycle (20 ns) significantly different from the feedback delay time (77 ns).
• * The network exhibited highly competitive performance on the Santa Fe chaotic time series prediction task.
• * Performance was robust across various tested operational conditions.

Conclusions:

• * Asynchronous reservoir computing based on semiconductor lasers is feasible and offers hardware implementation flexibility.
• * This approach achieves performance comparable to, or exceeding, synchronous systems for chaotic time series prediction.
• * The findings pave the way for more adaptable and efficient neuromorphic computing architectures.