<< Click to Display Table of Contents >> Navigation:  Using SMILE Wrappers > Learning

Learning in SMILE supports two primary tasks:

•Structure learning: Create a new network from a dataset.

•Parameter learning: Refine the parameters (conditional probability tables) of an existing network.

SMILE also supports network validation, which is commonly performed after learning to evaluate the quality and reliability of the resulting network.

Both learning and validation use data stored in a DataSet object. The DataSet class provides a table-like container for variable values used during training and evaluation. . Its content is typically loaded from a comma-separated values (CSV) file or a similar tabular format. A DataSet can also be populated programmatically, for example by iterating over database query results. This is an application-specific task handled by the user’s program.