Draft:Perception Anomaly Detection Algorithm
![]() | Draft article not currently submitted for review.
This is a draft Articles for creation (AfC) submission. It is not currently pending review. While there are no deadlines, abandoned drafts may be deleted after six months. To edit the draft click on the "Edit" tab at the top of the window. To be accepted, a draft should:
It is strongly discouraged to write about yourself, your business or employer. If you do so, you must declare it. Where to get help
How to improve a draft
You can also browse Wikipedia:Featured articles and Wikipedia:Good articles to find examples of Wikipedia's best writing on topics similar to your proposed article. Improving your odds of a speedy review To improve your odds of a faster review, tag your draft with relevant WikiProject tags using the button below. This will let reviewers know a new draft has been submitted in their area of interest. For instance, if you wrote about a female astronomer, you would want to add the Biography, Astronomy, and Women scientists tags. Editor resources
Last edited by 81.77.87.245 (talk | contribs) 3 months ago. (Update) |
The Perception algorithm[1][2][3] is an unsupervised parameter free machine learning technique designed to detect anomalies in univariate and multivariate data. The algorithm was inspired by elements of human perception, and builds on the idea that any event that is unexpected to occur, but does occur, is considered an anomaly with respect to some chosen measure. The algorithm is fast and efficient without the user having to supply any data specific parameters, as it adapts to the data distribution.
Algorithm
[edit]Given a list of integers Q = [...] it is assumed that they have been generated by a stream of indicators (1's and 0's) over a window of size L that is simple taken to be the largest integer in the list, we compute the following expectation of the number of n-tuples,
where S=sum(Q), W=number of windows (i.e. number of integers in list Q), and n is the value to be tested from the integers list Q. Any value of n for which the expectation of its occurrence is < 1, but it has occurred, is considered by definition an anomaly.
Applications
[edit]The algorithm has been used in three known production use cases:
- Detect malicious activities in corporate networks by application of the method to particular features of interest.
- Identify start and end points of promotional activity using electronic point of sales (EPOS) data from retailers, where unusually large drops or jumps in price over some minimum period of time indicate start and end of promotions.
- Detect unexpected number of support tickets being produced by a node in telecommunication applications to identify malfunctioning or compromised nodes.
References
[edit]- ^ Mohammad, Nassir (2022-05-13), "Anomaly Detection using Principles of Human Perception", arXiv:2103.12323 [cs.CR]
- ^ Mohammad, Nassir (2024-08-16), M-Nassir/perception, retrieved 2024-08-16
- ^ perception-nassir: A method for detecting anomalies in univariate and multivariate data, retrieved 2024-08-16