Fraud is Everywhere

The Oxford dictionary defines fraud as, "wrongful or criminal deception intended to result in financial or personal gain". Fraud can be the use of stolen credit, debit, fuel or gift cards, opening a bank account through identity theft, concealing the source of illegally or criminally received money, making a wrongful or exaggerated claim for medical expenses through insurance or social security claims, etc.  Fraud can be found in:

• on-line shopping
• computer gaming
• telecommunications
• banking
• social security
• tax evasion
• customs fraud
• insurance claims
• payments and financial services
• etc.

  The list of crimes and their targets is disappointingly long.

A Challenge

The detection of fraud is a complex scientific and business challenge. 

The datasets are large and therefore need to be properly sampled, as it typically becomes computationally impractical to use the natural population.  Real world data is noisy, unbalanced, computationally expensive to store and maintain and highly dimensional.  Such data is skewed, has uneven distributions and contains a mixture of symbolic and continuous variables.  The dataset is sparse - there is a large quantity of transactional data (typically millions of transactions a day), which contain only a small number of frauds.  Often researchers are unable to report on real fraud datasets due to their perceived sensitivity with many published papers using either synthetic or small datasets; where there is no reason to believe that the conclusions drawn will hold true when they are scaled-up to large, real-world unbalanced datasets.  In a live environment, fraud needs to be detected on-the-fly using a standard hardware environment that can be easily maintained and to help enforcement, a clear case needs to be presented for the reasons behind the fraud alert.

Our Mission

Existing technology approaches to Fraud Management are not keeping pace with the growing levels of all types fraud.  It is the mission of this Fraud Task Force to help close the gap between today's current practice and this ongoing crisis through the use of new computational intelligence technologies.

Let's Reduce Fraud Crime Now

The Fraud Task Force will explore the advancements and applications of computational intelligencefor practical fraud detection applications.

• neural computing
• evolutionary computation
• neuro-symbolic
•  AI
•  other emerging learning or optimisation techniques

To achieve this, the Fraud Task Force will facilitate the collaboration between commercial anti-fraud firms, academic researchers, enforcement and government to explore the impact of current research in a practical environment. 

The lack of large, real datasets in the field of fraud for the academic community hampers the research into practical new approaches to detection.  It is an aim of the Fraud Task Force to facilitate cooperation between researchers and the commercial world to make such data sets publically available with permission, where these have been sufficiently obfuscated to overcome security and other data protection concerns. 

This will lead to the advancement of fraud detection knowledge and the promotion of applied research. 

Activities

• Dissemination of scientific research applicable to fraud detection to take place through channels such as:
  
  • A Special Session will be proposed for the IEEE IJCNN-2012 (International Joint Conference on Neural Networks)
  • The IEEE Computational Intelligence Society (CIS) conference

  • Other established international conferences, reputable journals and subject-specific workshops organised by the Chair in close collaboration with key figures worldwide

  • A business and academic joint seminar will be organised with emphasis on the practical and deployable application of computational intelligence technologies – which will provide a mechanism to gain cooperation between researchers working in the field and importantly industry leaders
    
• Creation of public domain fraud datasets
  
•  Creation of agreed metrics for measurement and benchmark comparison of efficacy of algorithms; e.g. sensitivity, specificity, likihood ratio, ROC, comprehensibility, accuracy, paired t-test, F-measure, Wilcoxon signed-rank test, plus those specific to fraud detection, etc.
  
•  Potential "Competition" on public domain dataset to find best detector (classifier) and best explanation for fraud and to encourage cross-discipline research and debate
  
•  General interest articles will be published in commercial and non-scientific media to raise awareness. 
•  Collaboration with industry sector associations
  
•  Fraud Task Force membership drive through contacting key business leaders, Government and researchers

Members

If you think you can help then please join us! All Fraud Task Force members are welcome and are considered "thought leaders" in their industry sector.  You will be asked to comment on aspects of your profession and to share your perspective and vision to help tackle fraud.

Your Team

Chairman	Nick F Ryman-Tubb is the Director of Industrial Liaison, Digital Ecosystems Group and a Researcher at the Department of Computing, University of Surrey.  In 1987, he founded UK Company, Neural Technologies Ltd.  In 1993, Ryman-Tubb was chosen as the principal technical advisor to the UK Government £6m Neural Technology Transfer Program and during this time he authored a number of publications . He developed a new approach to data-mining using neural computers – reported to be the first such application in the UK. Ryman-Tubb worked on NASA and Government projects as well as practical applications:  in particular Fraud Management Systems, risk analysis and credit scoring.  He developed a proprietary neural computing system the Advanced Modular Adaptive Network (AMAN), a hybrid system that was then used in a range of anti-fraud products;  headlines include: "BoS Banking Direct Recruits Neural Fraud Detective", "New Software Teaches itself to Detect Credit Card Fraud", "Neural Computing System Designed to Stop Fraudsters".  These approaches included knowledge and rule-extraction algorithms that won a UK Government award for innovation in 1995.  Today, anti-fraud solutions based on these techniques developed by Ryman-Tubb, scrutinise billions of transactions every day and provides protection for 1-in-7 of the world's mobile telephone subscribers from fraudsters.  Stepping down from Neural Technologies in 2000, he joined City University London as an Honorary Visiting Fellow and developed a novel approach (SOAR) to neural-symbolic fraud rule extraction.  In 2010, Ryman-Tubb moved to the University of Surrey where he continues to research new intelligent approaches to fraud detection and to promote research and commercial collaborations.
Vice-Chair	Dr. Mark Goldspink, is at Logica UK.  Dr Mark Goldspink has spent the last 10 years working in the payment sector including authorisation, settlement, analytical fraud prevention and the issuing of fuel cards, with large global blue chip clients, including Shell, BP, Chevron, WalMart, Tesco’s, 02, T-Mobile, Discover Card and Travelex.  Previous to this he spent 12 years in the global downstream (marketing) oil sector with a background in Retail and Commercial and Industrial fuel and non fuel sales.  Currently Mark is working at Logica with responsibilities for developing their largest payment managed service contract with Shell and converting this into an industry service. Before that he was the managing director at Retail Decisions plc which during his tenure was voted 29th in the 2009 Sunday Times Tech List for the fastest growing companies.  Previous to this Mark worked in the oil industry where he demonstrated significant commercial and financial acumen by turning the 4th largest UK retail fuel business £1.5bn unit back into profit, £75m within a 12 month period. During his time he developed global sales and marketing strategies for one of the largest downstream oil company joint ventures in the world and consulted on business development projects in Thailand, Korea, South Africa, Singapore, Caribbean, Brazil and for the European Bank (EBRD) on a $200m Bulgarian downstream retail investment.
Vice-Chair	Ian Horobin is CEO and founder of Omnicision Limited.  Ian is an experienced professional with 16+ years experience working in the design and delivery of risk management, AML, Terrorist Financing and Fraud systems within Europe and North America. Ian has extensive exposure within this market and has published papers, spoken at numerous conferences and represented the UK at a FATF meeting on the use of technology to address Money Laundering. Ian founded Omnicision in 2006 to deliver product neutral consultancy services to companies exposed to financial crime and is currently working with a number of FI's across Europe to deliver strategic Financial Crime Solutions and to provide tuning and assurance assessments of Sanctions filters.

Contact

Contact the Chairman click to email to request more details and to join this important task force.

 

---