TOP > 外国特許検索 > Binary data classification method, binary data classification device, computer program, and storage medium

Binary data classification method, binary data classification device, computer program, and storage medium

外国特許コード F110003832
整理番号 I022P008WO
掲載日 2011年7月6日
出願国 アメリカ合衆国
出願番号 22353007
公報番号 20090049085
公報番号 8019762
出願日 平成19年2月2日(2007.2.2)
公報発行日 平成21年2月19日(2009.2.19)
公報発行日 平成23年9月13日(2011.9.13)
国際出願番号 JP2007051807
国際公開番号 WO2007088973
国際出願日 平成19年2月2日(2007.2.2)
国際公開日 平成19年8月9日(2007.8.9)
優先権データ
  • 特願2006-026330 (2006.2.2) JP
  • 2007JP051807 (2007.2.2) WO
発明の名称 (英語) Binary data classification method, binary data classification device, computer program, and storage medium
発明の概要(英語) An information processing apparatus 100 for realizing a binary data classification method of the present invention includes a CPU for computing a column vector a that has at least a quarter of its components equal to zero, which satisfies diag(y)Dna>0, where a represents a column vector having a coefficient of each term of the set polynomial function as an element, Dn represents a matrix determined on the basis of a combination of the values taken by the respective terms, and y represents a row vector having as an element the value of a class to which binary data in which a value of each element is 1 or −1 should be classified when the binary data is given, and thus classifies the data of an object of classification, which is inputted through a keyboard, in accordance with a set polynomial function.
従来技術、競合技術の概要(英語) BACKGROUND ART
A data classification method for classifying data in a database having a large quantity of information into a plurality of classes is becoming an essential art for information processing in recent years.
As for classification of certain data, the data can rarely be classified clearly, so that a method is proposed in which learning is carried out by using data for learning, which has been accurately classified in advance, and classification is carried out on the basis of the learning result.
For example, supervised learning to automatically learn how to classify the data from data for learning, of which correct answer has been known in advance, a learning method using a kernel function such as a support vector machine has been known (for example, refer to Patent Document 1).
[Patent Document 1] Japanese Patent Application Laid-Open No. 2000-293502

特許請求の範囲(英語) [claim1]
1. A data classification method comprising: setting a function to define binary data made of n pieces (n is an integer number not less than 2) of elements as an input value; computing the value of the function by substituting the set function with respective elements of the given binary data; and carrying out classification of the binary data on the basis of the value of the computed function;
wherein, a value of each element is 1 or -1; the binary data inputted within an information processing apparatus is defined to be an object of classification; a polynomial function is set by a CPU as computing means within the information processing apparatus so as to classify the binary data into two classes; and a column vector a which satisfies diag(y)Dna>0 is computed by using the CPU, where a represents a column vector having a coefficient of each term of the set polynomial function as an element, Dn represents a matrix determined on the basis of a combination of the values taken by the respective terms, and y represents a row vector having as an element the value of a class to which binary data should be classified when the binary data is given so that a polynomial function having terms in a number fewer than 3 * 2n-2 is obtained.
[claim2]
2. A data classification device: for setting a function to define a binary data made of n pieces (n is an integer number not less than 2) of elements as an input value; computing the value of the function by substituting the set function with respective elements of the given binary data; and carrying out classification of the binary data on the basis of the value of the computed function; comprising:
means for accepting binary data in which a value of each element is 1 or -1;
means for setting a polynomial function by using a CPU as a computing device so as to classify the binary data into two classes; and
means for computing a column vector a which satisfies diag(y)Dna>0 by using the CPU, where a represents a column vector having a coefficient of each term of the set polynomial function as an element, Dn represents a matrix determined on the basis of a combination of the values taken by the respective terms is Dn, and y represents a row vector having as an element the value of a class to which binary data should be classified when the binary data is given,
wherein, by these means, a polynomial function having terms in a number fewer than 3 * 2n-2 is obtained.
[claim3]
3. A computer readable storage medium storing a non-transitory computer program which allows a computer to set a function to define a binary data made of n pieces (n is an integer number not less than 2) of elements as an input value; and to compute the value of the function by substituting the set function with the respective elements of the given binary data; and to carry out classification of the binary data on the basis of the value of the computed function;
wherein the storage medium stores a computer program, comprising a step of:
allowing a computer to compute a column vector a which satisfies diag(y)Dna>0 by using a CPU as a computing device within the computer, where a represents a column vector having a coefficient of each term of the set polynomial function as an element, Dn represents a matrix determined on the basis of a combination of the values taken by the respective terms, and y represents a row vector having as an element the value a class to which binary data whose elements have a value of 1 or -1 should be classified when the binary data is inputted in the computer, and
allowing the computer to set a polynomial function having terms in a number fewer than 3 * 2n-2, by using the CPU so as to classification the binary data into two classes by using the column vector a, which is computed by the above step.
  • 発明者/出願人(英語)
  • OZTOP ERHAN
  • JAPAN SCIENCE AND TECHNOLOGY AGENCY
国際特許分類(IPC)
米国特許分類/主・副
  • 707/737
  • 707/776
ライセンスをご希望の方、特許の内容に興味を持たれた方は、問合せボタンを押してください。

PAGE TOP

close
close
close
close
close
close