TWed Discussion: Using Information Centrality for anomaly detection in large networks

Printer-friendly version

Website: TWed Discussion: Using Information Centrality for anomaly detection in large networks
When: April 6 2016
Where: Room 1140, Winslow Building, RPI Campus, Troy, NY, US
Description:
TWed Talk: Wed, 06 Apr (7p-8p, 1140 Winslow)
TITLE: "Using Information Centrality for anomaly detection in large networks"
LEADER: Nidhi Rastogi
VIDEO: TWed video streams
EVENT: tbd
KEYWORDS: Information centrality, graph theory, cyber attacks

Please join us this WEDS, 06 Apr (7p, 1140 Winslow) as PhD student Nidhi Rastogi leads us in an update on her interesting research applying the concept of "information centrality" to the problem of detecting cyber attacks in large networks.

DESCRIPTION: Large-scale collection of data, while a boon to modern data collection and analysis techniques, also poses a huge challenge in removing noise from the more useful data. Researchers have approached this problem through various means - machine learning of different data types, clustering those that should or shouldn't be present in the packet layer and labeling them according to their characteristics. This complements other existing techniques of reducing the amount of data to be analyzed making anomaly detection a much faster process. The goal remains to minimize data collection without compromising the quality of data collected. However, approaches to this end differ in means, application and kind of data to be analyzed - is it stored or real time. This research takes this effort forward by taking a graph theoretic approach to large networks that need to be analyzed for identifying anomalies for a systemic detection of cyber attacks. It works by identifying specific nodes, known as node centralities that can monitor anomalies effectively and rapidly. Node centralities can be measured through various means and can depend on the type of flow or transfer across a network. Information centrality is used to sparsify the graph and compare various network fingerprints to identify anomalous behavior, thus proving that IC can be used for this purpose as well.

BIO: Ms. Nidhi Rastogi is pursuing PhD in Computer Science at RPI with research in Cyber Security of distributed systems. Dr. Jim Hendler is her adviser. Currently she is exploring malware propagation in distributed systems when it is under a targeted attack. Another topic she is pursuing is identifying anomalous activities by using graph analytics fundamentals. Prior to joining RPI, Nidhi worked in the industry for four years in security and remote management of wireless devices on cellular networks at Verizon Wireless, NJ, and security of devices operating on the smart grid at Logic Technology Inc., Schenectady for the client GE Global Research Center, Niskayuna. Nidhi also has a Masters in Computer Science from the University of Cincinnati, OH where her thesis research was on improving security protocols in heterogeneous wireless networks during vertical handoff.

====
TWed Logistics (Spring):
  • TWed schedule
  • SPRING 2016 TWeds WILL BE 6p WEDS
  • Snacks will be provided for TWed Talks
  • Live video streams of TWed Talks are now available via YouTube
  • An archive of past TWed Talks is also available on ustream.
  • TWed Talks from previous terms are archived; topical archive coming soon!
====
About TWed:
  • "TWed" is the Tetherless World Educational Series
  • "TWed Talks" are informal overview talks and tutorials on topics of interest to the Tetherless World community. TWed gives members of the lab the chance to share tools and expertise. TWed talks are not lectures; they are expected to be highly interactive and fun. TWed leaders are encouraged to include live "hack" activities in their session plans.

Deadlines

Publications

Attending