Can I read Shannon Entropy Helps Optimize the Performance of a Frequency-Multiplexed Extreme Learning Machine on EtoBox?

Shannon Entropy Helps Optimize the Performance of a Frequency-Multiplexed Extreme Learning Machine by Zajnulina, Marina is a scholarly article available to read on EtoBox.

What is Shannon Entropy Helps Optimize the Performance of a Frequency-Multiplexed Extreme Learning Machine about?

Knowing the dynamics of neuromorphic photonic schemes would allow their optimization for controlled data-processing capability at possibly minimized energy consumption levels. In nonlinear substrates such as optical fibers or semiconductors, these dynamics can widely vary depending on each optical input encoded with information to process and are in general difficult to estimate. Thus, other approaches are required to optimize the schemes. Using a frequency-multiplexed fiber-based Extreme Learning Machine as an example for a classification task, I use Shannon entropy for optical power and introduce Shannon entropy for optical phase and spectrum, all averaged only over a small subset of the dataset to process. I show that the maximum and upper moderate optical power and phase entropies relate to the best data-processing capability of the Machine and, thus, can be used to find optimal system parameters such as fiber length, input power, and group-velocity dispersion in a more time-efficient manner than running numerical simulations emulating the scheme. Shannon entropy of the spectrum provides the information for what parameter space broadest possible frequency combs can be expected

Author

Zajnulina, Marina

Published

2024

Language

EN