Friday 9 January 2015

Top 10 Technologies For Stopping Fraud

Payments experts expect security budgets to increase this year – as well as the breaches. Forrester predicts that 60 percent of enterprises will discover a breach in 2015. But as fraud prevention solutions become obsolete by the day, says Brighterion, more efficient technologies that use real-time artificial intelligence are crucial. In a recent whitepaper, the company reveals how organizations can better defend themselves in the war against fraud using 10 of these top technologies as weapons, including Smart Agents, real-time profiling, and even “Fuzzy Logic.”

TODAY’S FRAUD SOLUTIONS
The solutions that exist today around fraud detection and prevention are built from technologies like Business Rules, Neural Networks, Data Mining and Statistics.

1) Business Rules require experts to divide the considered population into categories, and then write rules for identifying fraud within each category.

The Downside: First, there are many different types of individual behavior within each category, so broadly applying rules in these cases would result in poor detection. Second, rules are “hard-wired” into the system – they can’t adapt to ever-changing fraud schemes or data shifts.

2) Data Mining: This technology creates a “decision tree” from historical data. It applies the same logic to all entities, even though each entity may have a unique activity pattern.
The Downside: Decision trees are stagnant while fraud schemes continue to evolve – they fail to detect new types of fraud.

3) Neural Networks learn from historical data for future classification.
The Downside: As with Data Mining, these cannot detect any new fraud schemes (that arise daily) and apply the same logic to all entities, which inhibits the ability to detect new fraud schemes.

HOW SMART-AGENT TECHNOLOGY WORKS
According to Brighterion, Smart-Agent technology learns the specific behaviors of an entity (card, phone, computer, individual, etc.) to create virtual representations of each individual entity. That virtual agent will continuously learn from activities and behaviors, using its knowledge to detect any abnormal activity in real-time.

Instead of being programmed to anticipate every possible scenario or relying on structured, broad rules, Brighterion’s Smart-Agent technology iPrevent “learns, predicts, infers and makes intelligent real-time decisions” specific to each cardholder, merchant, ATM and more, the company says.
Screen Shot 2015-01-07 at 4.11.14 PM
The real-time decision making feature is what has been missing from the previous solutions. Another facet that Smart-Agents cover (and older technologies lack) is point of compromise detection. Not only do Smart-Agents identify fraud in the traditional sense, but they also identify phony merchants.

THE TOP 10 ARTIFICIAL INTELLIGENCE TECHNOLOGIES TO COMBAT FRAUD
Smart-Agent technology alone, like any other technology, is not the “silver bullet” in fraud protection and prevention. Brighterion therefore combines 10 different artificial intelligence technologies with its solution in order to bring the most comprehensive approach to fraud prevention:

  • Smart-Agents: Learns the specific behaviors of a given entity.
  • Constraint Programming: Constraints added to the system that defines what is or isn’t allowed in a particular outcome.
  • Neural Networks: Proprietary neural networks translate any database to neurons without user intervention.
  • Business Rules Management System: Embeds patented rule-engine algorithm that can accept an unlimited amount of rules.
  • Fuzzy Logic: Traditional and classical logic typically categorizes information into binary patterns like “black/white, yes/no, true/false, day/night.” Fuzzy Logic handles uncertainty in data.
  • Real-Time Profiling: Uses real-time profiling to evaluate behaviors of each individual entity.
  • Data Mining: Extracts knowledge from data.
  • Long-Term Profiling: Builds a profile for each individual entity (merchants, cardholders, ATMS, etc.) to understand their normal behavior.
  • Case Based Reasoning: Uses past experiences or cases to prevent and detect fraud.
  • Adaptive Learning: Allows models to learn and make effective changes at runtime.

UNDERSTANDING THE iPREVENT SOLUTION
In a world where data breaches are happening regularly – and the price of recovering from a data breach is steadily rising – timing is vital when it comes to anti-fraud tactics. The iPrevent solution automatically creates a predictive model based on a customer’s historical information – it then immediately is capable of preventing fraudulent transactions, says Brighterion.

Source:
pymnts.com

No comments: