FEATURE-QUANTITY EXTRACTING APPARATUS

• US 20150019171A1
• Filed: 06/06/2014
• Published: 01/15/2015
• Est. Priority Date: 07/11/2013
• Status: Active Grant

• Current Assignee: Casio Computer Company Limited
• Original Assignee: Casio Computer Company Limited
• Inventors: Yahata, Takashi, Kanke, Takao
• Granted Patent: US 10,143,406 B2
• Time in Patent Office: Days
• US Class Current: 702/190
• CPC Class Codes: A61B 5/1112 Global tracking of patients, e.g. by using GPS
• CPC Class Codes: A61B 5/1116 Determining posture transitions
• CPC Class Codes: A61B 5/1118 Determining activity level
• CPC Class Codes: A61B 5/1123 Discriminating type of movement, e.g. walking or running A61B5/1116, A61B5/112 take precedence

Abstract

A feature-quantity extracting apparatus is provided, which can calculate a proper feature quantity, by performing a simple calculating operation. The apparatus is provided with a code-string acquiring unit for acquiring code strings for every given period from a series of input data, wherein the code string is an arrangement of codes and the code is given to a value of each piece of input data, a code-string pattern frequency counting unit for counting the number of code-string patterns for every code-string pattern among the code strings acquired by the code-string acquiring unit, wherein the code-string pattern represents a code-string whose codes are arranged in accordance with a given order, and a feature-quantity outputting unit for outputting the number of code-string patterns for every code-string pattern counted by the code-string pattern frequency counting unit as a feature quantity of the series of input data.

Claims

1. A feature-quantity extracting apparatus comprising:

a code-string acquiring unit which acquires code strings for every given period from a series of input data, wherein the code string is an arrangement of codes and the code is given to a value of each piece of input data;

a code-string pattern frequency counting unit which counts the number of code-string patterns for every code-string pattern among the code strings acquired by the code-string acquiring unit, wherein the code-string pattern represents a code-string whose codes are arranged in accordance with a given order; and

a feature-quantity outputting unit which outputs the number of code-string patterns for every code-string pattern counted by the code-string pattern frequency counting unit as a feature quantity of the series of input data.

2. The feature-quantity extracting apparatus according to claim 1, wherein

the series of input data is time series data which is output from a sensor within a predetermined interval.

3. The feature-quantity extracting apparatus according to claim 2, wherein

the code included in the code string is information corresponding to a positive or negative symbol given to a difference value between separate pieces of data among the time series data which are output from the sensor respectively at different times.

4. The feature-quantity extracting apparatus according to claim 3, wherein

the code string comprises the given number of codes selected from a predetermined time from plural codes that vary time sequentially and are apposed in a line.

5. The feature-quantity extracting apparatus according to claim 4, wherein

the code-string acquiring unit acquires the code strings respectively at temporally deviated times from the plural codes varying time sequentially and being apposed in a line.

6. The feature-quantity extracting apparatus according to claim 1, wherein

the feature quantity is calculated form input data of known learning quantity for a classification class to use the input data, thereby generating a classifier for classifying a classification class of said input data.

7. The feature-quantity extracting apparatus according to claim 6, wherein

the feature quantity is calculated form input data of unknown quantity for the classification class to use the classifier upon receipt of said input data to classify a classification class of said input data.

8. A method of extracting a feature quantity comprising:

a step of acquiring code strings for every given period from a series of input data, wherein the code string is an arrangement of codes and the code is given to a value of each piece of input data;

a step of counting the number of code-string patterns for every code-string pattern among the acquired code strings, wherein the code-string pattern represents a code-string whose codes are arranged in accordance with a given order;

a step of outputting the number of code-string patterns counted for every code-string pattern as a feature quantity of the series of input data.

9. A non-transitory computer-readable storage medium with an executable program stored thereon, wherein a computer is mounted on a feature-quantity extracting apparatus and the program instructs a computer to perform the following steps:

a step of acquiring code strings for every given period from a series of input data, wherein the code string is an arrangement of codes and the code is given to a value of each piece of input data;

a step of counting the number of code-string patterns for every code-string pattern among the acquired code strings, wherein the code-string pattern represents a code-string whose codes are arranged in accordance with a given order;

a step of outputting the number of code-string patterns counted for every code-string pattern as a feature quantity of the series of input data.