Communications of the ACM
Photon-Based Processing Units Enable More Complex Machine Learning
The photonic tensor core performs vector-matrix multiplications by utilizing the efficient interaction of light at different wavelengths with multistate photonic phase change memories.
Credit: Mario Miscuglio
George Washington University (GWU) researchers proposed a new machine learning approach that induces a photonic tensor core to conduct multiplications of matrices in parallel, boosting the speed and efficiency of current deep learning paradigms.
A photonic tensor processing unit (TPU) stores and processes data in parallel, while an electro-optical interconnect enables efficient reading and writing of optical memory, allowing the TPU to interface with other architectures.
GWU's Mario Miscuglio said, "Integrated photonic platforms that integrate efficient optical memory can obtain the same operations as a tensor processing unit, but they consume a fraction of the power and have higher throughput and, when opportunely trained, can be used for performing inference at the speed of light."
From AIP Publishing
View Full Article
Abstracts Copyright © 2020 SmithBucklin, Washington, DC, USA
No entries found