Compact, Large-Scale Photonic Neurons by Modulation-and-Weight Microring Resonators

Fabrication imperfections, spatial constraints, and prohibitive costs collectively impede the scalability of neuromorphic photonics. In this work, we introduce a large-scale, compact photonic neuron in which each microring performs modulation and weighting simultaneously. This dual functionality is realized by leveraging both the carrier effect and thermal tunability, thereby merging modulation and weighting to conserve on-chip area, enhancing tuning efficiency, and capitalizing on wavelength-division multiplexing (WDM) for scalable implementations. In addition, we investigated a range of configurations for the proposed neuron to better tailor its behavior to various computational tasks. To illustrate the adaptability of the system's tasks, we explore both spatial and temporal domains, highlighting its versatility through two representative tasks: image processing and, for the first time, financial time series analysis, which represents a promising new frontier for neuromorphic photonics. These findings underscore the considerable promise of photonic computing in addressing a breadth of real-world challenges, particularly under escalating demands for both scalability and flexibility.