Search for: All records

As deep neural network (DNN) models grow ever-larger, they can achieve higher accuracy and solve more complex problems. This trend has been enabled by an increase in available compute power; however, efforts to continue to scale electronic processors are impeded by the costs of communication, thermal management, power delivery and clocking. To improve scalability, we propose a digital optical neural network (DONN) with intralayer optical interconnects and reconfigurable input values. The path-length-independence of optical energy consumption enables information locality between a transmitter and a large number of arbitrarily arranged receivers, which allows greater flexibility in architecture design to circumvent scaling limitations. In a proof-of-concept experiment, we demonstrate optical multicast in the classification of 500 MNIST images with a 3-layer, fully-connected network. We also analyze the energy consumption of the DONN and find that digital optical data transfer is beneficial over electronics when the spacing of computational units is on the order of$$>10\,\upmu $$$>10\mu$m.

 

We propose a digital incoherent optical neural network architecture using the passive data routing and copying capabilities of optics for artificial neural network acceleration. We demonstrate a proof-of-concept experiment and analyze optimal use cases. 

Optical approaches to AI acceleration have gained intense interest recently due to the potentially breakthrough advantages of photonics: high bandwidth, low power consumption, and efficient data movement. We overview leading photonic AI platforms based on beamsplitter mesh networks, weight banks, and photoelectric multiplication. While the theoretical performance can be orders of magnitude beyond current state of the art, practical issues of chip area, input / output, and crosstalk paint a more nuanced near-term picture of photonic AI acceleration. Both fundamental and near-term limitations to energy efficiency are addressed, and bandwidth limitations due to temporal crosstalk are analyzed.