Respective positive and negative example correct rate setting-based controllable confidence machine algorithm

Abstract

The invention belongs to the machine learning field and provides a respective positive and negative example correct rate setting-based controllable confidence machine algorithm. The respective positive and negative example correct rate setting-based controllable confidence machine algorithm includes the following steps that: a binary classifier is trained according to a sample train set, and classification is performed on the train set according to the binary classifier, and a classification result is converted into an output value; a preset equidistant step length is gradually increased from an original point, and a positive example correct rate is calculated and is compared with a preset correct rate, so that a positive example threshold value t1 can be obtained, and a negative example correct rate is calculated and is compared with a preset negative example correct rate, so that a negative example threshold value -t2 can be obtained, and a threshold value range (-t2,t1) can be formed according to the positive example threshold value t1 and the negative example threshold value -t2; and classification results of unknown samples are distributed according to the threshold value range. The respective positive and negative example correct rate setting-based controllable confidence machine algorithm provided by the technical scheme of the invention has the advantages of control precision and flexible control.

Description

technical field [0001] The invention belongs to the field of machine learning, and in particular relates to a controllable confidence machine algorithm based on separately setting the correct rate of positive and negative examples. Background technique [0002] Confidence machine is to provide a credible judgment on the learning results in the process of machine learning or to perform preset classification processing on the learning results. Trusting machines has important practical significance in high-risk applications such as medical diagnosis. Confidence machine is a branch of machine learning that has not been studied for a long time. There are not many theoretical foundations and methods for realizing confidence machine learning. There are methods of directly constructing confidence, and indirect methods of constructing confidence. Preset classification processing can be performed to exclude low-confidence parts, thereby improving the reliability of the remaining part...

AI Technical Summary

Problems solved by technology

[0007] The purpose of the embodiments of the present invention is to provide a controllable confidence machine algorithm based on separately setting the correct rate of positive and negative cases, which solves the problems of insufficient confidence control accuracy and insufficient flexibility in the prior art

Images