Building the information kernel and the problem of recognition

TL;DR

The paper discusses creating a new representation of information called an information kernel, emphasizing importance ranking within data to improve AI and information theory applications.

Contribution

It introduces the concept of an information kernel that prioritizes essential information, aiming to enhance data processing and recognition tasks.

Findings

Proposes a method to rank information importance within data.

Highlights the potential of information kernels in AI recognition.

Suggests reduction of data size without significant loss of relevant information.

Abstract

At this point in time there is a need for a new representation of different information, to identify and organize descending its characteristics. Today, science is a powerful tool for the description of reality - the numbers. Why the most important property of numbers. Suppose we have a number 0.2351734, it is clear that the figures are there in order of importance. If necessary, we can round the number up to some value, eg 0.235. Arguably, the 0,235 - the most important information of 0.2351734. Thus, we can reduce the size of numbers is not losing much with the accuracy. Clearly, if learning to provide a graphical or audio information kernel, we can provide the most relevant information, discarding the rest. Introduction of various kinds of information in an information kernel, is an important task, to solve many problems in artificial intelligence and information theory.