Embodiment of Learning in Electro-Optical Signal Processors
This enables electro-optical analog computers to embody their own training process for more difficult tasks, representing an incremental advance in photonic computing.
The authors tackled the problem of training electro-optical analog computers by physically implementing the backpropagation algorithm on a delay-coupled architecture, resulting in a considerable decrease in error rate on three benchmark tasks compared to when backpropagation was not used.
Delay-coupled electro-optical systems have received much attention for their dynamical properties and their potential use in signal processing. In particular it has recently been demonstrated, using the artificial intelligence algorithm known as reservoir computing, that photonic implementations of such systems solve complex tasks such as speech recognition. Here we show how the backpropagation algorithm can be physically implemented on the same electro-optical delay-coupled architecture used for computation with only minor changes to the original design. We find that, compared when the backpropagation algorithm is not used, the error rate of the resulting computing device, evaluated on three benchmark tasks, decreases considerably. This demonstrates that electro-optical analog computers can embody a large part of their own training process, allowing them to be applied to new, more difficult tasks.