Optics Letters, Год журнала: 2025, Номер 50(4), С. 1176 - 1176

Опубликована: Янв. 6, 2025

A synaptic device is fundamental for memory and learning functions of neural networks. In this work, we demonstrated a GST (Ge2Sb2Te5)-based microring synapse with nonvolatile reconfigurable characteristics. The shows bidirectional weight modulation during unidirectional crystallization or amorphization GST, simulating long-term depression (LTD) potentiation (LTP). addition, adopted an anti-resonance pump scheme to reduce the fluctuations in energy coupled into less than 12%. This significantly enhances programming precision synapse, achieving 24 resolvable states over 15 cycles. work holds promise laying foundation novel, best our knowledge, photonic computing architectures provides possibilities implementation vision adaptive optical networks that rely on plasticity.

Язык: Английский