It is often a single connector that provides access to artificial intelligence
Boost Fraud Prevention with Machine Intelligence: PredictiveWorks. Aerospike Connector
Fraudsters do not wait for Rules
Experts predict online credit card fraud to soar to $32 billion in 2020. For banks, brokerage houses, and payment gateways, fraud prevention is vital.
Both false positives and false negatives can have huge business implications. Failure in these areas results in losses to the bottom line and brand damage.
Fraud prevention applications are real-time applications and latency, often less than a second, is key to success. The lower the latency, the more work can be done to test:
Was the card reported stolen? Is the charge over the limit? Is the transaction unusual? Was the request made online or in person? Is the device recognized?
Today’s fraud prevention most often apply tailored rules to decide whether a certain activity or an incoming ATM transaction is fraudulent. Companies know that pure rule based fraud detection is not enough to keep pace with the ever-changing fraud landscape:
They begin to invest huge amounts of money into artificial intelligence for more proactive fraud detection.
Low latency NoSQL database for Fraud Detection
Aerospike is a low latency NoSQL database for large scale datasets with a fraction of the infrastructure complexity and costs, compared to other NoSQL databases.
Aerospike powers real-time solutions in plenty of industries, including Advertising Technology, E-Commerce & Retail, Financial Services & Payments, Online Gaming or Telecommunications.
Different industries, but facing a common challenge: Fraud.
For many companies in these industries, Aerospike became their favorite low latency database to boost real-time fraud prevention applications.
Aerospike beyond Rules
Machine Intelligence to fight Fraudsters
From a machine intelligence perspective, fraud detection is usually framed as anomaly detection or classification problem.
Both approaches learn from huge amounts of historical data what is fraudulent, legit or normal and describe this knowledge in a data model. Then, such a model is used to decide e.g. whether a certain ATM transaction is anomalous, fraudulent or legit.
Nowadays, augmenting existing fraud detection solutions with machine intelligence is complex and often a longer-standing task before companies can benefit from trained fraudster models.
Aerospike meets Cloud Data Fusion
Google’s Cloud Data Fusion charged with 200+ plugins, with connectors to modern data sources and plugins for full-spectrum machine intelligence, offers a code-free alternative and helps companies to boost their fraud prevention solution with artificial intelligence fast:
For fraud prevention based on Aerospike, it is nothing more than selecting the right data connector and historical customer behavior and transaction data can be used to train accurate fraudster models without writing a line of code.
Model building and deployment are based on the same technical foundation:
A data model that has been accepted for fraud detection is directly available and lo longer has to be reimplemented for production.
This architecture sketch illustrates how popular setups for real-time fraud detection with Aerospike can be augmented with code-free machine intelligence.
In this sample approach, model building is enabled by selecting the Aerospike Plugins.
Model usage in real-time environments is made available by choosing the Kafka Plugins.
Today’s fraud detection is often based on rules to decide, but this is not enough to keep pace with the continuously changing fraud landscape.
So what do we need? Machine Intelligence to fight fraudsters.
PredictiveWorks. charges Google’s cloud data fusion not only with 200+ plugins, but also with the needed connector to the favorite low latency database of most industries, Aerospike.
There is nothing more to do than select the right data connector and access historical data to train accurate fraudster models. Without the need to write code.