Did you know?
This architecture was completely manually coded and optimized in Verilog Hardware Description Language (HDL) for a single Field Programmable Gate Array (FPGA) chip, thus it can be easily adopted in any lab for small-to-medium SNN sizes. The architecture is parametric in the number of neurons that can be simulated, with limitations imposed by the hardware only in terms of the possibility of fitting the SNN in a commercial FPGA. The long-term goal of the so developed architecture is to bi-directionally interface the SNN with a biological one
(wheat9) HDFS-8091: ACLStatus and XAttributes should be presented to INodeAttributesProvider before returning to client (asuresh) HDFS-8305: HDFS INotify: the destination field of RenameOp should always end with the file name (cmccabe) HDFS-7980. Add LazyPersist flag to INodeFile, save it in FsImage and edit logs. MiniDFSNNTopology#simpleFederatedTopology(int) always hardcode nameservice ID
We now use FindNeighbors to generate a neighbor graph and SNN graph, and this is stored in the object. FindClusters then uses the NN or SNN graph to find clusters. The save.SNN parameter is no longer needed in FindClusters