Fraud Detection
Fraud Detection

Fraud Detection

Fraud detection is a difficult task. The proportion of fraudulent electronic transactions to legitimate ones is very small (prevalence). With the rapid growth in the number of web sites, merchants, and mobile applications, the total effort required to manage and prevent fraud is increasing. One key problem is how to find a small number of fraudulent transactions among millions of legitimate ones without inconveniencing 'good customers' and without imposing unwieldy manual reviews on fraud analysts. One of the biggest concerns of banks, transaction processors, insurance companies, merchants and online stores (CNP) is the loss of confidence of the customer. Fraud has a deep adverse impact on customer retention and customer acquisition. preventing Mebone Fraud uses two approaches to detect fraud:
  • Machine learning techniques to find fraudulent operations.
  • Enabling the fraud analyst to construct rules for fraud detection based on his or her expertise using her/his business language.
The first approach finds fraud faster than the authorization system can. The techniques used fall in the field of artificial intelligence. The latter lets the analyst find fraud by mastering the rules and adapting them to find new fraud patterns.
But this is only a part of continuous process of fight against fraud
Diagram


 
Mebone® Fraud enables the possibility
to catch fraud in electronic transactions
even before the fraud is commited.
The Difference
Innovation

Innovation

There are different techniques used in fraud detection, from the application of Specific Rules to Artificial Intelligence Systems based on neural networks (ANN) or mathematical algorithms. Mebone® Fraud uses Rules and Predictive Algorithms inside a Machine Learning strategy.
The system provides several elements that make it particularly innovative,
  • Both Rules and Predictive detection strategies are built in natural or business language. There is no required technical specialist in order to build and run Rules or Predictive Models into Mebone® platform.
  • The predictive engine is able to learn continuously, without retraining processes, avoiding periods of degradation, stop and reset at specialized laboratory.
  • Artificial intelligence provided by the predictive engine is automatically readjusted to offer a set of the best models among all possible ones.
  • The information for optimizing predictive models is additionally available throughout the system for use by analysts of fraud in Rules building and the Fraud Prevention tasks
  • Detection strategies, both based on rules such as those run in the predictive engine, respond in real time to transactions analysis demand, allowing early decision making, critical risk management and financial damages avoiding.
  • It is a Big Data platform with capabilities for processing large volumes of data, extracting the maximum value to information.

Informacion

 
The combination of state of the art
techniques, makes Mebone® Fraud an
unique tool for fraud detection.
Transparency
Transparency

Transparency

High added value is that detection systems explain why a particular transaction is alerted. In the case of the rules, it is simple, as the rule itself explains the intended results. However, the explanation on Expert systems is not equally simple, and becomes opaque especially in the case of neural networks (black box).

Analytics. Machine Learning module of Mebone® Fraud, brings together a set of tools that provide detailed information of the power of discrimination (legal/ilegal) that has each of the values of each of the elements of the analyzed information regarding fraud detection.


Fraud Prevention. It is important an early fraud detection and identification of pattern's changes as well as the readjustment of the scan engines to those changes. Mebone® Fraud provides information about such changes as well as a systematic selection of the best predictive strategies for fraud detection. The system offers at the same time all the information of changes to risk analysts, which allows to deploy a fraud prevention task.

 
Mebone® Fraud provides the information
that the analyst needs, in a practical,
simple way and with full transparency.
The Performance
The Performance

The Performance

The analysis of significant data for fraud detection involves reviewing large volumes of data. This type of analysis could consume minutes or even hours using traditional strategies. Mebone® Fraud makes use of specialty engines that optimize queries and associated calculations, exponentially increasing its performance. Both, Rules and predictive models, are able to respond in Real Time window (milliseconds).

Building detection strategies. Mebone® Fraud provides an intuitive graphical interface for both the generation and management of Rules as well as for modeling and building of predictive models. Risk analysts, specialized in these tasks, have their wizard that guides them on constructing rules, which significantly increase efficiency in construction and the immediate availability of fraud detection strategies.

Case Management. The system provides to the risk analysts all information associated with the case being reviewed, in order to manage alerts properly. The transparency of the system provides very useful data for managing the alert itself but is also an added value in building new rules and predictive models for detection.
The platform integrates with SMS gateways, email servers, Apps, including the ability to parameterize direct cardholder communication, taking in account the detection strategy, customer segment, alert type, severity, etc.

Mebone® Fraud provides detection rates and true positive ratios that certify their efficiency (depending on prevalence of fraud), while allowing to control on-line, the balance between the volume of alerts that the entity is able to review and the target of money recovery in each case.

Eficiencia

 
Highly optimized algorithms makes
Mebone® Fraud capable to analyze
transactions data in real time.
 
Read More

Fraud Detection

Fraud detection is a difficult task. The proportion of fraudulent electronic transactions to legitimate ones is very small (prevalence). With the rapid growth in the number of web sites, merchants, and mobile applications, the total effort required to manage and prevent fraud is increasing. One key problem is how to find a small number of fraudulent transactions among millions of legitimate ones without inconveniencing 'good customers' and without imposing unwieldy manual reviews on fraud analysts. One of the biggest concerns of banks, transaction processors, insurance companies, merchants and online stores (CNP) is the loss of confidence of the customer. Fraud has a deep adverse impact on customer retention and customer acquisition. preventing Mebone Fraud uses two approaches to detect fraud:
  • Machine learning techniques to find fraudulent operations.
  • Enabling the fraud analyst to construct rules for fraud detection based on his or her expertise using her/his business language.
The first approach finds fraud faster than the authorization system can. The techniques used fall in the field of artificial intelligence. The latter lets the analyst find fraud by mastering the rules and adapting them to find new fraud patterns.
But this is only a part of continuous process of fight against fraud
Diagram


Read More

Innovation

There are different techniques used in fraud detection, from the application of Specific Rules to Artificial Intelligence Systems based on neural networks (ANN) or mathematical algorithms. Mebone® Fraud uses Rules and Predictive Algorithms inside a Machine Learning strategy.
The system provides several elements that make it particularly innovative,
  • Both Rules and Predictive detection strategies are built in natural or business language. There is no required technical specialist in order to build and run Rules or Predictive Models into Mebone® platform.
  • The predictive engine is able to learn continuously, without retraining processes, avoiding periods of degradation, stop and reset at specialized laboratory.
  • Artificial intelligence provided by the predictive engine is automatically readjusted to offer a set of the best models among all possible ones.
  • The information for optimizing predictive models is additionally available throughout the system for use by analysts of fraud in Rules building and the Fraud Prevention tasks
  • Detection strategies, both based on rules such as those run in the predictive engine, respond in real time to transactions analysis demand, allowing early decision making, critical risk management and financial damages avoiding.
  • It is a Big Data platform with capabilities for processing large volumes of data, extracting the maximum value to information.

Informacion

Read More

Transparency

High added value is that detection systems explain why a particular transaction is alerted. In the case of the rules, it is simple, as the rule itself explains the intended results. However, the explanation on Expert systems is not equally simple, and becomes opaque especially in the case of neural networks (black box).

Analytics. Machine Learning module of Mebone® Fraud, brings together a set of tools that provide detailed information of the power of discrimination (legal/ilegal) that has each of the values of each of the elements of the analyzed information regarding fraud detection.


Fraud Prevention. It is important an early fraud detection and identification of pattern's changes as well as the readjustment of the scan engines to those changes. Mebone® Fraud provides information about such changes as well as a systematic selection of the best predictive strategies for fraud detection. The system offers at the same time all the information of changes to risk analysts, which allows to deploy a fraud prevention task.

Read More

The Performance

The analysis of significant data for fraud detection involves reviewing large volumes of data. This type of analysis could consume minutes or even hours using traditional strategies. Mebone® Fraud makes use of specialty engines that optimize queries and associated calculations, exponentially increasing its performance. Both, Rules and predictive models, are able to respond in Real Time window (milliseconds).

Building detection strategies. Mebone® Fraud provides an intuitive graphical interface for both the generation and management of Rules as well as for modeling and building of predictive models. Risk analysts, specialized in these tasks, have their wizard that guides them on constructing rules, which significantly increase efficiency in construction and the immediate availability of fraud detection strategies.

Case Management. The system provides to the risk analysts all information associated with the case being reviewed, in order to manage alerts properly. The transparency of the system provides very useful data for managing the alert itself but is also an added value in building new rules and predictive models for detection.
The platform integrates with SMS gateways, email servers, Apps, including the ability to parameterize direct cardholder communication, taking in account the detection strategy, customer segment, alert type, severity, etc.

Mebone® Fraud provides detection rates and true positive ratios that certify their efficiency (depending on prevalence of fraud), while allowing to control on-line, the balance between the volume of alerts that the entity is able to review and the target of money recovery in each case.

Eficiencia