A neural network is an interconnected assembly of simple processing elements, units or nodes whose functionality is loosely based on the animal neuron. The processing ability of the network is stored in the inter-unit connection strengths, or weights, obtained by a process of adaption to, or learning from a set of training patterns.
The artificial equivalents of biological neurons are the nodes or units. Synapses are modeled by a single number or weight so that each input is multiplied by a weight. The weighted signals are summed together to supply a node activation. The activation is then compared with a threshold; if the activation exceeds the threshold, the unit produces a high-valued output, otherwise it outputs zero.
Neural Networks are often used for statistical analysis, classification and data modeling.
Limitations of Neural Networks
Neural Networks are algorithmic systems which use historical data to identify trends, clusters, and patterns. Unsupervised neural networks (clustering) are inefficient and inadequate. Supervised learning are limited by their training, i.e. they can reliably recognize only the kind of information on which they were trained.
IT Researches Neural Networks Library
iPrevent's proprietary Neural Networks technology can translate any database to neurons without user intervention and has significantly accelerated the speed of convergence as compared to typical Neural Networks algorithms such as Back Propagation. iPrevent's Neural Networks are incremental and adaptive, allowing the size of the output classes to change dynamically. Additionally, in its expert mode, iPrevent provides a library of twelve different Neural Networks models for use in customization.
Case-Based Reasoning (CBR) uses past experiences (cases) to solve current problems and can be applied in:
- Process Control
- Advanced Manufacturing
Learning and Generalization in iPrevent's CBR
The inductive indexing capabilities in iPrevent's CBR provide major advantages over Neural Networks, Business Rules, and Data Mining by learning from a wider range of past experiences. iPrevent's CBR technology translates a database to cases without user intervention. The cases created are then used to classify the normal/abnormal behavior in Real Time.
Traditional and classical logic typically categorizes information into binary patterns such as: black/white, yes/no, true/false, or day/night. Fuzzy Logic handles uncertainty in data.
iPrevent's Fuzzy Logic technology automatically clusters the information into various risk categories and improves performance by decreasing sensitivity to noisy data or outliers.
iPrevent's Fuzzy Logic is integrated with Brighterion's Neural Networks, Genetic Algorithms, Business Rules, Constraint programming and Case-Based Reasoning. iPrevent is the only Business Rules Management System in the market that allows fraud experts to write powerful rules with the following syntax: When the number of cross border transactions is high and when the transaction is done late in the night then the transaction is suspicious.
Genetic Algorithms belong to the class of evolutionary algorithm which generates solutions to optimization problems using techniques inspired by natural evolution, such as inheritance, mutation, selection and crossover.
A population of chromosomes is created and evaluated by the cost function, with the "most fit" chromosomes being kept in the population while the "least fit" ones are discarded. The chromosomes are then paired so they can mate, this involves combining portions of each chromosome to produce new chromosomes. During the mating process random mutations are often used. The new chromosomes are evaluated by the cost function and the process iterates in order to get closer and closer to the best solution.
iPrevent's Genetic Algorithms are used in the fields/attributes selection and combined with iPrevent's Neural Networks in the task of weight as well as architecture optimization.
Business Rules Management System
BRMS or Business Rule Management System is a software system used to define, deploy, execute, monitor and maintain the variety and complexity of decision logic that is used by operational systems within an organization or enterprise. This logic, also referred to as business rules includes policies, requirements, and conditional statements that are used to determine the actions that take place in applications and systems.
IT Researches Business Rule Management Systems
IT Researches BRMS provides an intuitive web-based browser interface for writing, editing and testing of the Business Rules or meta rules without any programming skills. Brighterion BRMS gives control of business rules back to the business team
With Brighterion Business Rules Management Systems you can:
- Rapidly build rule-based applications and deploy them into almost any operating environment.
- Change rules at any time, without disrupting operations and without IT assistance.
- Develop a repository of rule sets for scoring, routing, decision making and data transformation.
- Use the rules in Real Time and/or Batch mode.
- Evaluate the quality of a rule or a set of rules using various thresholds.
- Create and manage multiple rule sets.
- Manage users, roles and privileges.
- Sort Business Rules for improved processing.
- Handle any kind of transaction, interact and import any external data.
- Integrate with industry standard fromat :TCP, ISO8583, EURONET, XML, Data Base, Text File, etc.
- Enable creation of rules via a text editor and graphical interface.
- Define and maintain categories or groups of rules as a single entity.
- Validate the syntax and the coherence of the rules.
- Provide a complete audit-trail of information and traceability for each rule.
- Use XML: Brighterion Business Rules can be saved and imported in an XML format.
- Scale (Superior Algorithm): Brighterion BRMS are embedded with a unique rule-engine algorithm, which is faster than RETE (the industry standard algorithm) and can accept an unlimited number of rules.
- 99.999% operational availability guaranteed...no software downtime.
- Benefit from the power of Fuzzy Logic: Brighterion BRMS is the only system on the market that allows fraud experts to write rules with a powerful syntax such as: When the number of cross border transactions is high and when the transaction is done very late in the night then...
With Brighterion BRMS, there is no limit to the number or kinds of rules that non-IT business users can dynamically implement and deploy without stopping the system.
Constraint programming is a programming paradigm wherein relations between variables are stated in the form of constraints. Constraints differ from the common primitives of imperative programming languages in that they do not specify a step or sequence of steps to execute, but rather the properties of a solution to be found. This makes constraint programming a form of declarative programming.
The CP paradigm consists of a set of variables, each of which have a set of possible values (their domain) and a set of constraints between the variables that specify which combinations of values are allowed and which are not.
Brighterion's constraint programming technology is a complete language that integrates the following concepts:
Variables: Real numbers, integers, enumerated, sets, matrices, vectors, intervals, fuzzy subsets and more.
Arithmetic Constraints: : =, +, -, *, /, /=, >, <, ?, ?, interval addition, interval subtraction, interval multiplication, interval division, max, min, intersection, union, exponential, modulo, logarithm and more.
Temporal Constraints: Temporal constraints including equal, nequal, before, after, meets, overlaps, starts, finishes and personal temporal operators such as disjoint, started-by, overlapped-by, met-by, finished-by and more.
Boolean Constraints: Or, and, not, xor, implication, equivalence
Symbolic Constraints: Inclusion, union, intersection, cardinality, belonging and more.
Fuzzy Constraints: To achieve the best possible solution in cases where no exact solution exists.
Brighterion's constraint programming technology relieves programmers of the burden of learning a new language. Most programmers can generate their first optimization program in less than one hour.
Data mining is the process of extracting knowledge from data by uncovering previously unknown useful information and relationships.
iPrevent's Data Mining algorithms (over a dozen different algorithms) also include the following tools:
- Data Quality: error correction, outlier detection, imputation of missing values, feature selection, incoherence correction, data preparation & enrichment algorithms, etc.
- Statistics: regression analysis, correlation, multiple comparison, CHAID, etc.
- Probabilistic Inference: Bayesian networks, graphical model, hierarchical and probabilistic cluster analysis, etc.
- Association rule learning
- Graphical Visualization
Text mining is the discovery of previously unknown information by automatically extracting information from unstructured or structured text files.
Brighterion Text Mining:
- Automatically identifies entities, relationships (link analysis), topics, categories and clusters.
- Goes far beyond screening for vocabulary and uses a proprietary Fuzzy Logic and thesaurus-indexing algorithms that can extract the meaning from documents.
- Supports many formats such as: HTML, XML, plain text, PDF, Microsoft Office, etc.
- Allows parallel processing of large amounts of data with high throughput.
iPrevent's Velocity Analyzer enables financial institutions to monitor a wide variety of customer and merchant data, such as production purchasing patterns, suspicious changes in activities, and number of transactions over a period of time. iPrevent Velocity Analyzer uses a powerful compression technique to save transactions for optimal scalability and performance.
Additionally, institutions can monitor:
- Payment method history and typical purchasing
- Patterns at the merchant's site
- E-mail address activity
- Ship-to/Bill-to activity
- Refund Watch, Manual T-Log
- Excessive Cash Back
- Decline Analyzer
- Excessive Failed Pre-Authorizations
- Unattended and Attended Transactions