Photonic extreme learning machines using event-based detection

Photonic extreme learning machines use random optical propagation, detection nonlinearity, and a trained linear readout for energy-efficient and scalable optical computing. However, conventional intensity readout with CCD or CMOS cameras constrains the dimensionality of the hidden representation space. Here, we experimentally replace intensity detection with an event-based camera, whose thresholded log-intensity response provides alternative pixel-wise hidden representations: first-event time, binary activation, and event count. In a nonlinear two-spiral classification task, we obtain accuracies of 93±3% with strong intrinsic generalization, and comparable ridge and pseudo-inverse performance indicating sample-limited effective dimensionality. Regression results reveal sensitivity to systematic representation drift, identifying stability requirements for future event-based PELMs. These results establish event-based detection as an alternative nonlinear optical-to-digital measurement that provides informative representations for the considered tasks while clarifying current limitations.