MIDAS: A New Approach with Real-Time Streaming Anomaly Detection in Dynamic Graphs

Image Source: https://github.com/bhatiasiddharth/MIDAS?fbclid=IwAR2omTgS8UTrjopjJ57-PQdHIHp8cFNOlKfD3csHTFEgTHMg-jyD0EGNr-s

While most of the Machine learning articles are focussed on self-driving cars, GAN, and Image recognition, there are some other important areas that AI researchers and data scientists are working on. This includes researches to solve anomaly detection, which helps in network security to preventing financial fraud protecting businesses, individuals, and online communities. To help improve anomaly detection, Siddharth Bhatia (Ph.D. candidate) and his team at the National University of Singapore, have developed MIDAS (Microcluster-Based Detector of Anomalies) in Edge Streams. MIDAS is a new approach to anomaly detection that outperforms baseline approaches both in speed and accuracy. What makes MIDAS different from other available tools is its ability to detect these anomalies in real-time at speed greater than existing state-of-the-art models.


  • Finds Anomalies in Dynamic/Time-Evolving Graphs
  • Detects Micro-cluster Anomalies (suddenly arriving groups of suspiciously similar edges e.g., DoS attack)
  • Theoretical Guarantees on False Positive Probability
  • Constant Memory (independent of graph size)
  • Constant Update Time (real-time anomaly detection to minimize harm)
  • Up to 48% more accurate and 644 times faster than the state of the art approaches

Introduction to TensorFlow for Artificial Intelligence, Machine Learning, and Deep Learning [Advertisement]

Microcluster-Based Detector of Anomalies in Edge Streams


Github: https://github.com/bhatiasiddharth/MIDAS?fbclid=IwAR2omTgS8UTrjopjJ57-PQdHIHp8cFNOlKfD3csHTFEgTHMg-jyD0EGNr-s

Paper: https://arxiv.org/pdf/1911.04464.pdf

Related Shopping (Affiliate links)


Please enter your comment!
Please enter your name here

This site uses Akismet to reduce spam. Learn how your comment data is processed.