Information Technology Reference
In-Depth Information
9. How is a model of subject knowledge built for machine learning?
In order to build a model of subject knowledge for machine learning without a teacher, the
following needs to be done:
1.
Divide structures of knowledge into variables that is into linguistic variables and
invariant variables namely, into a universal set.
2.
Invariant knowledge or universal set of knowledge is denoted by X.
3.
This knowledge is classified divided into classes or elements of the set and is build on
the horizontal. The resulting classes of invariant knowledge or elements of the set are
denoted by x 0.1 ... x0, 9. Invariant knowledge or the set of knowledge can be finite or
infinite according to the classification of the classes themselves.
4.
Variable, categorical knowledge or linguistic variable are located vertically and denoted
by U. They are also classified and denoted by 0.1; 0.2 and etc [3]. In work [Nordhausen
and Langley, 1990] it was noted that formation of categories - is the basis of a unified
theory of scientific research. Denoting classes and groups of the set, as well as
categorical knowledge of a linguistic variable in number, we can present this
knowledge with their properties. Each property takes a serial number. So the sentence
"The day was sunny" can be defined on a coordinate plane as points: xa, 1,1 y a, 1, x a, 5,
y, a 4 / 1 xa, 2 ya1 / 5. Each element of the set, for example, x0,1 is characterized by its
own linguistic variable - y0,. Numerical parameters x0,1,1 by the rules of Russian
grammar mean masculine noun, y0, 1 mean a single number, etc. Analogous to these
characters, a set of proposals on the basis of available knowledge can be constructed.
This property can be one of the justifications for machine learning without a teacher as
it was derived from an example of the natural language. Each class of knowledge or
element of the set is at the same time, a cluster of the structure of knowledge. They have
their own rules and laws. Structures of knowledge clusters can be combined, divided,
associated and canceled according to logical settings of rules of a linguistic variable.
Belonging of the linguistic variable into elements of the set or the logical structures of
clusters is determined by logical operations such as operation of substitution,
enrichment, identical and multiplicative operations. That is, in this case as the rules of
linguistic variable and the rules of elements of the set are in mobile motion all the time,
by combining according to the given settings around the logical structure of knowledge
or elements of the set become nanostructures of knowledge.
10. What innovations can bring this knowledge model in education?
1.
Knowledge is considered in the scheme of integrity. Since language is a means of
communication and expression of ideas, then, certainly, the conditions of integrity
scheme and logic of Zadeh can be in any scientific knowledge and the rules of logic of
Piaget and fuzzy logic can be applied to them.
2.
Structures of knowledge will be divided into invariant and variable or into syntactic
and semantic. Numeric designation of categorical and invariant properties of
knowledge makes it possible to build coordinates of knowledge on the basis of which
the process of knowledge construction will go.
3.
Operationality of thinking enables to collect structures of knowledge into clusters,
figure out their interrelationship and attitudes, classify them, enrich, or replace with the
other structures. These logical operations gather as a magnet around the structures of
Search WWH ::

Custom Search