![Institutional Repository [SANDBOX]](images/dias.logo.png "Institutional Repository [SANDBOX]"){kind=logo}
Institutional Repository [SANDBOX]
Technical University of Crete
Technical University of Crete
EN
|
EL
| URI: | http://purl.tuc.gr/dl/dias/511D251A-FCD6-4B17-95D6-748E0F42FB40 | ||
| Year | 2018 | ||
| Type of Item | Diploma Work | ||
| License |
|
||
| Bibliographic Citation | Emmanouil Fragiadoulakis, "ΑΜRules for fraud detection with Spark Streaming", Diploma Work, School of Electrical and Computer Engineering, Technical University of Crete, Chania, Greece, 2018 https://doi.org/10.26233/heallink.tuc.78971 | ||
| Appears in Collections |
In this day and age, an important part of our daily interaction with our electronic devices is on-line payments, which results in a great amount of transactions. In order to handle these transactions, to determine if they're fraudulent, we need an efficient, distributed and streamable machine learning algorithm, that can process big amount of incoming data and react to it instantly. Thus, we implemented the distributed Adaptive Model Rules on Spark Streaming, an extension of the Spark Core API which enables the development of scalable, fault-tolerant streaming applications. Adaptive Model Rules is an one-pass algorithm for training its model from streaming data and is robust to outliers and irrelevant features. The experimental results concluded, that there is a noticeable speedup from Vertical Adaptive Model Rules to Hybrid Adaptive Model Rules at the cost of reduced accuracy.
Copyright © DIAS 2013
