Implementing a neural network algorithm on a neurosynaptic substrate based on metadata associated with the neural network algorithm
AssigneeInternational Business Machines Corporation
InventorsArnon Amir, Rathinakumar Appuswamy, Pallab Datta, Myron D. Flickner, Paul A. Merolla, Dharmendra S. Modha, Benjamin G. Shaw
Life of this patent
Estimating the term…
Abstract
One embodiment of the invention provides a system for mapping a neural network onto a neurosynaptic substrate. The system comprises a metadata analysis unit for analyzing metadata information associated with one or more portions of an adjacency matrix representation of the neural network, and a mapping unit for mapping the one or more portions of the matrix representation onto the neurosynaptic substrate based on the metadata information.
Claims17
A non-transitory computer program product, the computer program product comprising a computer-readable storage medium having program code embodied therewith, the program code being executable by a computer to:
- receive metadata information associated with adjacency matrix representation of a neural network, wherein the metadata information is indicative of one or more portions of the adjacency matrix representation to map onto a neurosynaptic substrate and further indicative of one or more neurosynaptic substrate constraints;
- and
- map the one or more portions of the adjacency matrix representation onto the neurosynaptic substrate by:
- selecting at least one program implementation from a library based on the metadata information;
- and
- programming the neurosynaptic substrate based on the at least one program implementation selected, wherein the neurosynaptic substrate programmed satisfies the one or more neurosynaptic substrate constraints.
Dependencies are inferred from the claim text.