|
|
Core Technology
CTGi Script Writer
 Click to enlarge.
Write high performance scripts capable of billions of transactions per month, including the Managing of XML Validations, the validation of multiple service order types, and the validation, routing and processing of voice and data transactions.
Intelligent Agents
Patent Pending High-Speed A.I. Technology
The methods and formulas employed by the system were developed over two years of R&D. They take full advantage of 14 years of experience with pattern analysis, mathematical modeling, and artificial intelligence.
A feedback arrangement monitors the success and failure of the agents and rewards/punishes the agents, thereby enabling the program to learn. This improves the system performance based on past and current experience, and enables the system to continually self adapt.
Other agents loop back on the agents themselves, looking for groups of agents that are more accurate taken in combination - this performs a similar function to but is much faster than a neural net. Additional enhancements improve and speed the learning capabilities of the system. LeMarck, Darwin and RV are learning modes designed to leverage and circumvent the strengths and limits of self-learning A.I. systems. Of particular importance and uniqueness is the speed in which this self-learning artificial intelligence is implemented … the system is fast enough to make decisions in real time.
LeMarck
Rewards and punishes agents that voted for and against winning and losing trades. Keeps the updated agent settings if they produce a superior result as compared to all prior runs. Named for the scientist who proposed that offspring inherit the learned behaviors of their parents.
Darwin
Randomly changes agent settings, executes a replay, and keeps the altered settings if they produce a superior result as compared to prior runs. Darwin is particularly useful in breaking through dead ends or plateaus that LeMarck encounters. Named for the scientist who proposed natural selection.
High Speed Distributed Database
Mechanically, the database and learning sub-systems utilizes multiple computers networked together over a local area network using a high speed messaging protocol custom designed for this application, which is able to distribute the workload of decision making, correlation and self-learning. This patent pending technology allows us to process and correlate alarms, call records, SS7 packets and other network data in real time, and act on the results.
|
|