| PRTools contents |
SETBATCH
W = SETBATCH(W,FLAG,BATCHSIZE,OBJSIZE)
W = W*SETBATCH([],FLAG,BATCHSIZE,OBJSIZE)
| Input | |
| W | Mapping |
| FLAG | Switches batch processing on (TRUE, default) or off (FALSE) |
| BATCHSIZE | Number of objects per batch (default 1000) |
| OBJSIZE | Number of objects above which batch processing is enabled (default BATCHSIZE) |
| Output | |
| W | Mapping |
Mappings might be processed in batch mode in case of large datasets or datafiles. This may solve problems in computing A*W (A is a datase/file) by executing the result in batches if intermediate memory demand is too high. In case the final dataset is too large this will not be of use as it should fit in memory anyway.
For some mappings batch processing is not possible as the processing of objects is not fully independent. For that reason it is skipped during training of mappings and classifiers.
| PRTools contents |