Analog dual classifier via a time-modulated neuromorphic metasurface

A neuromorphic metasurface embodies mechanical intelligence by realizing physical neural architectures, enabling computational capabilities within an elastic substrate with minimal energy requirements. However, well-established constraints on wave propagation in finite media have limited such systems to a single task. This work introduces a wave-based dual classifier that utilizes a single neuromorphic metasurface to concurrently execute two completely independent classification problems, via dynamic modulation of embedded waveguides. These results point to paradigms in wave-based computing systems that have thus far been elusive.