Abstract
Extreme Learning Machines (ELMs) are a versatile Machine Learning (ML) algorithm that features as the main advantage the possibility of a seamless implementation with physical systems. Yet, despite the success of the physical implementations of ELMs, there is still a lack of fundamental understanding in regard to their optical implementations. In this context, this work makes use of an optical complex media and wavefront shaping techniques to implement a versatile optical ELM playground to gain a deeper insight into these machines. In particular, we present experimental evidences on the correlation between the effective dimensionality of the hidden space and its generalization capability, thus bringing the inner workings of optical ELMs under a new light and opening paths toward future technological implementations of similar principles.

Abstract
The growing demand for multiparameter sensors includes compact devices accompanied by simple calibration processes to distinguish the outputs from each other. This paper evaluates a scheme to determine multiple parameters of a medium using localized surface plasmon resonances (SPR) excited on a Dshaped photonic crystal fiber (PCF) partially covered by two gold layers of different thicknesses. We demonstrate that the proposed sensing platform, once customized to characterize the possible dispersive profiles of the refractive index of the analyte, also allows interrogating the temperature of a sample from a linear relationship. Since the plasmonic resonances are excited at separated and low crosstalk spectral channels, different sensing responses can be obtained simultaneously in the same location of the D-shaped PCF. These features turn out the SPR sensor a suitable tool for simultaneous monitoring of optical dispersion and temperature. © 2023 SBMO/SBMag.