本計畫擬使用本實驗室所提出的新型光學神經網路計算系統，並使用奈米光學元件的耦合器與相位調制器，達到提供神經網路非線性函數與優化的功能，以建立積體化的神經網路光學計算晶片。本計畫擬變化各式拓樸連接形態的神經網路，以優化其光學計算之特性。並擬將光學計算系統製作於silicon-on-insulator(SOI)晶片上。我們亦將類神經網路應用，放入全光學人工智慧電腦中進行計算。以此類神經網路奈米晶片所進行之光學計算，在未來可提供大數據等課題一個體積小、速度快的計算平台。 ;We plan to use the optical neural networks proposed by our laboratory to develop an integrated optical neural network chip using nano-optics components. We will use the directional couplers and phase modulators to provide the non-linear function for the neurons and to optimize the neural networks. We will fabricate the device on silicon-on-insulator chip. We will apply the applications of artificial intelligence on the chip. This optical neural network chip could provide an ultra-fast computation platform to the applications of big data.
