Databases Reference
In-Depth Information
FIGURE 8.2: Clustered DHGN configuration for a coarse-grained network.
Each DHGN node is capable of performing the entire subpattern recognition
processes.
such as grid and cloud computing, in which additional processing and storage
capacity can be made available.
An important benefit of having this DHGN cluster performed on a single
processing node is that it eliminates all of the communication actions in the
DHGN message-passing model for distributed systems. For each subpattern
recognition process, each node communicates the index generated to the SI
module. The absence of communication between nodes reduces the chances
of recognition failures attributed to transmission or communication errors.
When the clustered configuration is used to implement the DHGN, the DHGN
subnets are formed using the internal memory structure of a single node. An
associative array structure for each DHGN subnet was adopted. Table 8.2
shows the associative array structure for a DHGN subnet with 5-bit binary
subpatterns.
Communications between each GN memory structure in the DHGN subnet
are conducted using conventional value store/retrieve processes, which update
values using value assignment.
8.1.2 DHGN Commodity Grid Framework
Distributed pattern recognition provides an avenue for achieving Internet-
scale pattern recognition using a state-of-the-art data classifier for fast track-
