Non-Invasive Self-Adaptive Information States' Acquisition inside Dynamic Scattering Spaces.

Area of Science:

• Physics
• Information Theory
• Machine Learning

Background:

• Achieving high information acquisition efficiency is crucial for reaching measurement precision limits in scattering media.
• While engineered modes can maximize information states, they often require partial system intrusion.
• Non-invasive methods for dynamic scattering spaces are challenging due to complex mapping problems, especially for multi-target scenarios.

Purpose of the Study:

• To experimentally demonstrate the feasibility of non-invasive information state acquisition in dynamic scattering spaces.
• To address the challenges posed by the non-unique mapping problem in multi-target scenarios.
• To achieve the ultimate measurement precision limit without disturbing the system.

Main Methods:

• Introduction of a tandem-generated adversarial network framework.
• Utilizing the external scattering matrix of the system.
• Experimental validation in dynamic scattering environments.

Main Results:

• Successful non-invasive acquisition of information states in dynamic scattering spaces.
• Demonstration of efficient information state acquisition for multi-target scenarios.
• Achievement of the Fisher information limit solely through system's scattering matrix.

Conclusions:

• The developed adversarial network framework enables non-invasive information state acquisition in dynamic scattering systems.
• This approach overcomes previous limitations in multi-target scenarios and complex scattering environments.
• The findings offer new perspectives for high-precision measurements in dynamic complex systems.