Hypothesis space
The hypothesis space contains all possible hypotheses that are relevant to a mathematical problem. The dimensionality of the hypothesis space is often very large and each dimension can assume continuous and discrete values. Colloquially, one also says that all the attributes of the training examples span the hypothesis space.
The hypothesis space is closely linked to the version space.
example
The following example is about the soccer world championship in certain years, about their participants and their placements. A three-dimensional space is spanned:
- (France, 2nd place, 2006)
- (Germany, 1st place, 2014)
- (Austria, 3rd place, 1954)
The hypothesis space contains many more hypotheses. Some of these have already occurred, others can be considered false, and some have not yet occurred.
literature
- Tom M. Mitchell: Machine Learning , McGraw Hill, 1997. ISBN 0071154671