org.dllearner.core
Interface Oracle
public interface Oracle
An oracle can be used by a learning algorithm to interactively ask for the
classification of an individual. Note that an oracle can either be a user or
an automatic method, which means that the implementation of an oracle can
reach from simple checks to more complex user interaction processes.
Usually, an oracle should be instantiated by passing a learning problem
to its constructor.
- Author:
- Jens Lehmann
|
Method Summary |
List<Double> |
classifyIndividuals(List<Individual> individuals)
This method should be called by a learning algorithm if it wants a list of individuals
(including the special case of a single individual) to be classified by the oracle. |
classifyIndividuals
List<Double> classifyIndividuals(List<Individual> individuals)
- This method should be called by a learning algorithm if it wants a list of individuals
(including the special case of a single individual) to be classified by the oracle.
For each of the individuals, which are specified in the parameter, the oracle must
return a value in the obvious order. (The first element in the returned list classifies the first element
of the list of individuals, the second element in the returned list classifies the
second element of the individual list etc.) The following values should be used:
- -1.0: Indicates that the individual does not belong to the learned class.
- Values between -1.0 and 1.0 indicate the degree to which an individual belongs to the class.
- 1.0: The individual does belong to the learned class.
- -2.0: The oracle does not know how to classify the individual (e.g. no feedback given by a user).
Note that the most common case is that the individuals list contains a single element, which
is answered by 1.0 or -1.0 by the oracle. However, the oracle interface is designed with a
higher degree of flexibility in mind, which is required for some use cases.
- Parameters:
individuals - A list of individuals, which should be classified by the oracle.
- Returns:
- For each element of the list, a classification value as explained above is
returned.
DL-Learner is licenced under the terms of the GNU General Public License.
Copyright © 2007-2011 Jens Lehmann