The Hugin GUI supports two kinds of parametric learning; adaptive learning and EM (batch) learning.

Adaptive Learning
Adaptation (adaptive learning) is used when you wish to adapt all or some of your conditional probability tables in the knowledge base to a new set of data.

EM learning
EM (Estimation-Maximization) is a learning facility used for batch learning. Batch learning is "off-line" learning and is used when you want to generate your conditional probability tables in the knowledge base from data stored in a database.