6 Anomaly Detection
Learn how to detect rare cases in the data through anomaly detection - an unsupervised function.
Related Topics
See Also:
-
Campos, M.M., Milenova, B.L., Yarmus, J.S., "Creation and Deployment of Data Mining-Based Intrusion Detection Systems in Oracle Database 10g"
-
K. C. Gross, V. Bhardwaj and R. Bickford, "Proactive detection of software aging mechanisms in performance critical computers," 27th Annual NASA Goddard/IEEE Software Engineering Workshop, 2002. Proceedings., Greenbelt, MD, USA, 2002, pp. 17-23, doi: 10.1109/SEW.2002.1199445.
6.1 About Anomaly Detection
The goal of anomaly detection is to identify items, events, or observations that are unusual within data that is seemingly 'normal'. This data may consist of traditional enterprise data or Internet of Things (IoT) sensor data.
Anomaly detection is an important tool for detecting, for example, fraud, network intrusions, enterprise computing service interruptions, sensor time series prognostics, and other rare events that can have great significance but are hard to find. Anomaly detection can be used to solve problems like the following:
-
A law enforcement agency compiles data about illegal activities, but nothing about legitimate activities. How can a suspicious activity be flagged?
The law enforcement data is all of one class. There are no counter-examples.
-
An insurance agency processes millions of insurance claims, knowing that a very small number are fraudulent. How can the fraudulent claims be identified?
The claims data contains very few counter-examples. They are outliers.
-
An IT department encounters compute resource performance anomalies. How can such anomalies be detected along with their source causes, such as resource-contention issues and complex memory leaks?
The data contains sensor output from thousands of sensors.
-
An oil and gas enterprise or utility company requires proactive maintenance of business-critical assets, such as oil rigs or smart meters, to reduce operations and maintenance costs, improve up-time of revenue-generating assets, and improve safety margins for life-critical systems.
6.1.1 Anomaly Detection as a form of One-Class Classification
Learn about anomaly detection as one-class classification in training data.
When applied to traditional data, anomaly detection can be viewed as a form of one-class classification, because ideally only one class is represented in the training data. An anomaly detection model predicts whether a data point is typical for a given distribution or not. An atypical data point can be either an outlier or an example of a previously unseen class.
Normally, a classification model must be trained on data that includes both examples and counterexamples for each class so that the model can learn to distinguish between them. For example, a model that predicts the side effects of a medication must be trained on data that includes a wide range of responses to the medication.
A one-class classifier develops a profile that generally describes a typical case in the training data. Deviation from the profile is identified as an anomaly. One-class classifiers are sometimes referred to as positive security models, because they seek to identify "good" behaviors and assume that all other behaviors are bad.
In single-class data, all the cases have the same classification. Counterexamples, instances of another class, are hard to specify or expensive to collect. For instance, in text document classification, it is easy to classify a document under a given topic. However, the universe of documents outside of this topic can be very large and diverse. Thus, it is not feasible to specify other types of documents as counterexamples. Anomaly detection can be used to find unusual instances of a particular type of document.
Note:
Solving a one-class classification problem can be difficult. The accuracy of one-class classifiers cannot usually match the accuracy of standard classifiers built with meaningful counter examples.
The goal of this type of anomaly detection is to provide some useful information where no information was previously attainable. However, if there are enough of the "rare" cases so that stratified sampling produces a training set with enough counterexamples for a standard classification model, then the classification may be a better solution.
Related Topics
6.1.2 Anomaly Detection for Time Series Data
Learn about anomaly detection in IoT sensor data.
With the growing number of sensors in the internet of things, the ability to identify anomalous events among potentially thousands of sensors is essential. For example, in the early detection of anomalies in business-critical enterprise computing servers and software systems, storage systems, and networks.
Enterprises require high anomaly detection accuracy, which implies lower false-alarm probabilities, lower missed-alarm probabilities, and lower overhead compute cost. The ability to distinguish between a real problem and sensor malfunction can significantly reduce costs in problem solution.
Building a model involves supplying historical, error-free operating data from, for example, monitored equipment. The resulting model is used to score new sensor data, also referred to as the monitoring phase, to estimate the expected sensor values.
6.2 Anomaly Detection Algorithms
For anomaly detection, Oracle Machine Learning for SQL has the following algorithms.
- Multivariate state Estimation Technique - Sequential Probability Ratio Test (MSET-SPRT)
- One-Class Support Vector Machine (SVM)
Anomaly detection is a form of classification. When you create a model using the MSET-SPRT and One-Class SVM algorithms, specify the classification machine learning technique. These algorithms do not use a target.