In
computational learning theory
In computer science, computational learning theory (or just learning theory) is a subfield of artificial intelligence devoted to studying the design and analysis of machine learning algorithms.
Overview
Theoretical results in machine learning m ...
in
mathematics, a concept over a domain ''X'' is a total
Boolean function
In mathematics, a Boolean function is a function whose arguments and result assume values from a two-element set (usually , or ). Alternative names are switching function, used especially in older computer science literature, and truth function ...
over ''X''. A concept class is a class of concepts. Concept classes are a subject of
computational learning theory
In computer science, computational learning theory (or just learning theory) is a subfield of artificial intelligence devoted to studying the design and analysis of machine learning algorithms.
Overview
Theoretical results in machine learning m ...
.
Concept class terminology frequently appears in
model theory
In mathematical logic, model theory is the study of the relationship between formal theories (a collection of sentences in a formal language expressing statements about a mathematical structure), and their models (those structures in which the ...
associated with
probably approximately correct (PAC) learning.
[Chase, H., & Freitag, J. (2018). ''Model Theory and Machine Learning''. arXiv preprint arXiv:1801.06566]
In this setting, if one takes a set ''Y'' as a set of (classifier output) labels, and ''X'' is a set of examples, the map
, i.e. from examples to classifier labels (where
and where ''c'' is a subset of ''X''), ''c'' is then said to be a ''concept''. A ''concept class''
is then a collection of such concepts.
Given a class of concepts ''C'', a subclass ''D'' is reachable if there exists a sample ''s'' such that ''D'' contains exactly those concepts in ''C'' that are extensions to ''s''.
Not every subclass is reachable.
Background
A ''sample''
is a
partial function
In mathematics, a partial function from a set to a set is a function from a subset of (possibly itself) to . The subset , that is, the domain of viewed as a function, is called the domain of definition of . If equals , that is, if is ...
from
to
.
Identifying a concept with its characteristic function mapping
to
, it is a special case of a sample.
Two samples are ''consistent'' if they agree on the intersection of their domains.
A sample
''extends'' another sample
if the two are consistent and the domain of
is contained in the domain of
.
Examples
Suppose that
. Then:
* the subclass
is reachable with the sample
;
* the subclass
for
are reachable with a sample that maps the elements of
to zero;
* the subclass
, which consists of the singleton sets, is ''not'' reachable.
Applications
Let
be some concept class. For any concept
, we call this concept
''-good'' for a positive integer
if, for all
, at least
of the concepts in
agree with
on the classification of
.
The ''fingerprint dimension''
of the entire concept class
is the least positive integer
such that every reachable subclass
contains a concept that is
-good for it.
This quantity can be used to bound the minimum number of equivalence queries needed to learn a class of concepts according to the following
inequality:
.
References
Computational learning theory
{{Compu-AI-stub