ABNORMALITY DETECTION DEVICE, ABNORMALITY DETECTION PROGRAM, AND ABNORMALITY DETECTION METHOD
To provide means capable of presenting data having high possibility of abnormality and not appearing in a trend of feature quantity of teacher data.SOLUTION: An abnormality detection device 10 includes: a feature quantity generation unit 12 which generates a feature quantity vector from a log; an ab...
Gespeichert in:
1. Verfasser: | |
---|---|
Format: | Patent |
Sprache: | eng ; jpn |
Schlagworte: | |
Online-Zugang: | Volltext bestellen |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | To provide means capable of presenting data having high possibility of abnormality and not appearing in a trend of feature quantity of teacher data.SOLUTION: An abnormality detection device 10 includes: a feature quantity generation unit 12 which generates a feature quantity vector from a log; an abnormality learning and discrimination unit 13 which generates, by learning, an abnormality discriminator for discriminating abnormality of the feature quantity vector; a non-similarity learning and determination unit 14 which generates, by learning, a non-similarity calculator for determining non-similarity of the feature quantity vector; a feed-back priority label applying unit 15 which, in a log corresponding to the teacher data, outputs a value of an abnormal field as a key element, and applies a priority label to a feature quantity vector being determined to have high non-similarity of having the same element with the key element; and a feed-back candidate display unit 17 which preferentially displays feed-back candidate data to which the priority label is applied, and provides means capable of making feed-back of the truth/false of the data.SELECTED DRAWING: Figure 1
【課題】教師データの特徴量の傾向に現れず、異常可能性が高いデータを提示可能な手段を提供する。【解決手段】異常検知装置10は、ログから特徴量ベクトルを生成する特徴量生成部12と、特徴量ベクトルの異常を判別する異常判別器を学習により生成する異常学習/判別部13と、特徴量ベクトルの非類似度を判定する非類似度算出器を学習により生成する非類似度学習/判定部14と、教師データに対応するログにおいて、異常なフィールドの値をキー要素として出力し、キー要素と同一の要素を持つ非類似度が高いと判定された特徴量ベクトルに優先ラベルを付与するフィードバック優先ラベル付与部15と、優先ラベルが付与されたフィードバック候補データを優先的に表示し、当該データの正誤をフィードバック可能な手段を提供するフィードバック候補表示部17と、を有する。【選択図】図1 |
---|