Researchers have introduced a technique to use optics in probabilistic computing. In their work, they demonstrated that there are nonlinear optical phenomena that are highly suitable for resolving a graphical probabilistic model. The graphene based thin films for optical computing were created in the University of Eastern Finland, in Professor Yuri Svirko’s nanocarbon laboratory.
"Graphical probabilistic models are commonly used when in case of a large number of complex interacting datapoints. These models can be utilised, for instance, in machine vision, artificial intelligence, machine learning, speech recognition and computational biology. Processing of a large number of complex interacting datapoints requires efficient computers, while optically the solution could be obtained more naturally. By the presented optical techniques, the computing could be done faster and more efficiently than by those conventional manners. The optical computing was done by graphene like materials, which have recently shown great potential in optics," says Tommi Kaplas, a senior researcher who currently works in Vilnius, at the Center for Physical Sciences and Technology.
The research was conducted in collaboration with the University of Arizona, the University of Eastern Finland and Georgia Institute of Technology. The results of the study were published in the esteemed Nature Communications journal.
For further information, please contact:
Tommi Kaplas, tommi.kaplas@ftmc.lt
Original article:
Masoud Babaeian, Pierre-A. Blanche, Robert A. Norwood, Tommi Kaplas, Patrick Keiffer, Yuri Svirko, Taylor G. Allen, Vincent W. Chen, San-Hui Chi, Joseph W. Perry, Seth R. Marder, Mark A. Neifeld & N. Peyghambarian
Nonlinear optical components for all-optical probabilistic graphical model. Nature Communications volume 9, Article number: 2128 (2018), DOI: 10.1038/s41467-018-04578-x.
