Guiding principle of reservoir computing based on "small-world" network

Area of Science:

• Computational neuroscience
• Network science

Background:

• Reservoir computing, a framework for recurrent neural networks, offers simplified training.
• Optimal reservoir construction methodology remains underdeveloped.
• "Small-world" networks exhibit real-world network properties like high clustering and short path lengths.

Purpose of the Study:

• To provide a systematic approach for synthesizing optimal reservoirs.
• To leverage the controllable parameters of small-world networks for reservoir design.

Main Methods:

• Utilizing the controllable parameters of small-world networks.
• Developing a guiding principle for reservoir synthesis.

Main Results:

• A validated methodology for constructing desired reservoirs.
• Demonstrated effectiveness on classification and prediction tasks.

Conclusions:

• Small-world network properties can guide the systematic synthesis of optimal reservoirs.
• The proposed methodology is effective for diverse computational tasks.