Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Automatic Determination of Recognition Error Rates

IP.com Disclosure Number: IPCOM000090847D
Original Publication Date: 1969-Jul-01
Included in the Prior Art Database: 2005-Mar-05
Document File: 2 page(s) / 39K

Publishing Venue

IBM

Related People

Chow, CK: AUTHOR

Abstract

This apparatus for testing pattern recognition systems calculates automatically, to a close approximation, the recognition error rate that can be expected from the particular recognition system being tested. Such is effected without actually counting or explicitly identifying the errors.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 100% of the total text.

Page 1 of 2

Automatic Determination of Recognition Error Rates

This apparatus for testing pattern recognition systems calculates automatically, to a close approximation, the recognition error rate that can be expected from the particular recognition system being tested. Such is effected without actually counting or explicitly identifying the errors.

Each recognition system has a reject threshold t which can be adjusted by increments to different levels t(1), t(2), etc. The reject rate R of the system, which is the number of times that the system rejects sample patterns of a given set that fail to match the true patterns, usually changes as the reject threshold setting t is changed. The error rate for each setting can be predicted with reasonable accuracy from the change in value of the reject rate between adjacent settings, using the following recursive equation in which E(t(k)) represents the calculated error rate at the reject threshold t(k).

(Image Omitted)

In this system, a change in the threshold setting t(i) of the recognition system being tested causes a change in the reject rate R(ti), from which the error rate E for the new setting is automatically computed using the algorithm.

1

Page 2 of 2

2

[This page contains 3 pictures or other non-text objects]