Skip to content
/ RAMP Public

[code] "RAMP: Real-Time Anomaly Detection in Scientific Workflows" by J. Dinal Herath, Changxin Bai, Guanhua Yan, Ping Yang, Shiyong Lu. In: IEEE International Conference on Big Data (2019).

Notifications You must be signed in to change notification settings

dherath/RAMP

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

27 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

RAMP: Real-Time Anomaly Detection in Scientific Workflows

This repository includes a python implementation of RAMP (Real-Time Aggregated Matrix Profile), a machine learning model originally developed for real-time anomaly detection in scientific workflows. Please refer the paper for details on algorithms. The code includes an illustrative example for a non-interleaved scientific workflow dataset, generated from DATAVIEW workflow management system with synthetic anomalies of two types. The system logs were parsed into a multidimensional time series with three dimensions as shown in image below (the first three subfigures).

Comparing the truth labels (final subfigure) with the RAMP result for the anomaly scores clearly shows RAMP identifies all the anomaly instances, where the peaks greater than the threshold indicate anomalies. Additionally, RAMP is able to identify what features may have contributed to the anomaly instance (figure before last) where the two anomaly types can be clearly distuingished. Anomaly type-1 only causes changes in dimensions 1 & 3 (blue and green), where as anomaly type-2 causes changes in all three feature dimensions. Executing the code run_ramp.py will generate and save the figures for the example dataset.

time-series tresult

requirements: python3, numpy

@inproceedings{herath2019ramp,
  title={RAMP: Real-Time Anomaly Detection in Scientific Workflows},
  author={Herath, J Dinal and Bai, Changxin and Yan, Guanhua and Yang, Ping and Lu, Shiyong},
  booktitle={2019 IEEE International Conference on Big Data (Big Data)},
  pages={1367--1374},
  year={2019},
  organization={IEEE}
}

About

[code] "RAMP: Real-Time Anomaly Detection in Scientific Workflows" by J. Dinal Herath, Changxin Bai, Guanhua Yan, Ping Yang, Shiyong Lu. In: IEEE International Conference on Big Data (2019).

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages