Researcher: Gavin McWilliams
Research Institution: Queens University Belfast
Overview:
Gavin McWilliams from Queens University Belfast is seeking industry partners to help support further use of YAGADA in a broad range of applications, which will provide opportunity for refinement and case studies for evaluation.
Product:
• YAGADA is a graph-based anomaly detection algorithm;
• It allows data owners to condense large diverse data sets and expose anomalies that are infeasible to discover in a manual analysis;
• If the source data can be expressed as a graph (most data sets can be expressed graphically in some form) then YAGADA can deliver world-class anomaly detection performance; and
• Where data is already held in structured relational formats supporting tools can generate graphical versions ready for YAGADA execution.
Market:
• The discovery of financial fraud in large complex databases of financial transactions;
• The detection of medical anomalies in large datasets acquired through medical sensing;
• Early trend detection in social media content;
• Monitoring of access control systems deployed to physically secure large facilities; and
• Detection of attacks or malicious activities on communications networks.
Benefits:
• YAGADA improves on competing anomaly detection methods;
• It provides a solution for anomaly detection in both numerical and structural data – in reality most data has numerical and structural properties when modelled as a graph and traditional approaches have failed to deliver adequate anomaly detection in numerical data;
• It outperforms existing solutions in formal evaluations showing superior false-positive / true-positive performance; and
• YAGADA is currently being optimised for inline use allowing end users to apply YAGADA to time-constrained applications and detect critical anomalies that need immediate attention.
Partners:
YAGADA consists of a toolkit implementation that end users can readily lift and integrate into their existing data management solutions. Industry partners are sought to help support further use of YAGADA in a broad range of applications, which will provide opportunity for refinement and case studies for evaluation.
Specifically, YAGADA team are seeking partners to:
• Provide imaginative use case scenarios where YAGADA can be deployed and demonstrate commercial advantage;
• Source datasets where YAGADA can be openly evaluated to demonstrate performance and viability for use as a commercial capability; and
• Partner in the pursuit of new funding opportunities to further refine YAGADA and establish end-user awareness.
Contact Details:
Email Gavin McWilliams at g.mcwilliams@qub.ac.uk