Abnormal intelligence data recognition machine learning method

Abstract

The invention discloses an abnormal intelligence data recognition machine learning method, belongs to the field of intelligence data processing and mainly solves the problem that existing abnormal intelligence data recognition requires a lot of repeated manual debugging in an actual engineering application and data are difficult to directly apply. The method comprises steps as follows: firstly, collecting historical intelligence data of an information source, and performing manual analysis and study on abnormal data to form an original database; setting intelligence data training sample vectorcomposition, and generating an abnormal intelligence data recognition training dataset through feature extraction; further fitting probability distribution of the intelligence data, generating the best threshold probability in combination with label dataset iteration, and generating an abnormal intelligence data recognition model. According to a system, the abnormal intelligence data recognitionmodel is automatically trained and generated, a lot of manual debugging operation of model parameters is completely avoided, and the system has the advantages of high model generation speed and good practical effect.

Description

technical field [0001] The invention belongs to the field of intelligence data processing, relates to machine learning generation of abnormal intelligence data discriminant, and is suitable for intelligence preprocessing links. Background technique [0002] Anomalous intelligence data is data that deviates so far from other observations in intelligence preprocessing that it can be considered to have been generated by a different mechanism. The existence of abnormal intelligence data will affect the processing of intelligence data and reduce the accuracy and reliability of intelligence. Therefore, the identification and detection of abnormal intelligence data is the premise of intelligence data processing. At present, the commonly used abnormal intelligence data identification methods are mainly based on least squares estimation, polynomial filtering and smoothing differential technology, and Kalman real-time forecast and control technology based on observation model and sta...

AI Technical Summary

Problems solved by technology

[0003] The purpose of the present invention is to propose a machine learning method for abnormal intelligence data identification, aiming to solve the problem that the existing abnormal intelligence data identification method needs to be manually modified and debugged repeatedly using measured data in practical applications

Images