Meeting/Event Information

Mid-MO AGA Chapter presents, on behalf of the National AGA Webinar series...

Graph Modeling and Analytics for Fraud Detection:

Addressing the "Needle in a Haystack" Problem

May 10 | 1:00–2:50 PM ET | 2 CPEs | FOS: AUDG

Speaker: Andrea Fletcher, Executive Director of the Digital Services Centers for Medicare & Medicaid Services

Cost:  $30/member, $40/non-member

REGISTRATION IS LIMITED TO FIRST 30 PEOPLE!!

Graph Modeling and Analytics has emerged as an important tool for fraud detection. Fraud detection is a quintessential "needle in a haystack" problem, where there is a very small amount of known fraud and very large amount of unknown cases.

Current fraud detection efforts may be insufficient for the following reasons:

• The growing sophistication of fraudulent activities creates complex, high-dimensional, non-linear patterns.
• Rules based approaches do not scale well and produce high false positive rates.
• Fraudulent behavior is often deeply hidden behind seemingly innocuous behavior.
• If fraud monitoring is based on historical patterns of behavior, it will fail to identify a new fraudulent behavior until it is too late to act.

During this webinar, we will discover and discuss:

• How to use knowledge graphs to capture financial transactions, where a single account is represented as a vertex and a single transaction between two accounts is represented as an edge.
• Utilizing graph analytics to investigate the complex connections between individuals, accounts, companies, and locations, mapping of thousands of relationships (edges) between thousands of entities (nodes).
• A key concept is the "connectedness" between nodes. Connectedness can reveal relationships between nodes including:
  • Closeness - identify which nodes are most important to other nodes within a graph
  • Connected components - indication of relationships amount groups of nodes
  • Community detection - detecting groups that are densely connected internally but loosely connected externally
  • Ranking - establishing the trustworthiness of the nodes relative to the trustworthiness of similar nodes

REGISTRATION:  The Mid-MO AGA Chapter has already purchased the webinar through National.  If you would like to attend, you will need to register through the Mid-MO AGA Chapter website.  After registration, you will be provided with an access code to register with National, who will then send you the webinar link and materials.