Enhancing the Resolution of Local Near-Field Probing Measurements With Machine Learning

In this purely numerical work, we discuss the use of machine learning (ML) techniques to improve the resolution of local near-field probing (LNFP) measurements when the probe used in LNFP is larger than the device being studied. The study demonstrates that through the implementation of ML algorithms, it is possible to achieve a
$\lambda /10$
spatial resolution even with probes that are a few wavelengths wide, while maintaining a maximum relative error of less than 3%. The investigation further reveals that fully connected neural networks (FCNNs) exhibit superior accuracy compared to linear regression when dealing with limited training datasets. Conversely, for larger training datasets, it is unnecessary to construct and train neural networks, as linear regressions prove to be both sufficient and efficient. These findings establish the potential of employing similar ML approaches to enhance the resolution of measurements obtained from diverse experimental setups.