TinMan AI Builder's patents-pending IDE (integrated development environment) provides a complete suite of break-through, user-friendly tools to Design, Train, Simulate and Package an artificial intelligence engine for deployment in a host application.
Types of host applications may be: Military / Defense intelligence and unmanned vehicle applications, Computer Games, Robotics projects and autonomous behavior, Medical diagnostics/therapy recommendations, Customer analysis and prediction, Law Enforcement / Criminal analysis, Banking / investment / Fraud analysis and prevention, Aviation and Flight simulation, and any other target application that requires an autonomous artificial intelligence engine for making decisions based on a dynamic stream of data.
Overview of the key steps to designing and deploying an AI system:
- Design - Layout the Structure of the AI System
TinMan AI Builder and Integrated Development Environment [IDE] provides a best-of-breed combination of several forms of logic, improving flexibility, accuracy and reliability of the AI systems produced. The IDE allows the creation of autonomous AI cores through familiar drag and drop operation for the Inputs to a system, the outputs from a system and the neural components that provide the logical linkage between the inputs and outputs. The AI systems produced with TinMan AI Builder can receive input from many inputs simultaneously, and accurately produce and manage many output actions simultaneously. The models themselves can be interconnected, allowing both finite (single/exclusive) and multi-state management (simultaneous state execution) with near limitless depth and breadth descendent system processing.
- Train - Embed Expert Knowledge into the AI System
The embedding of expert knowledge into an AI system in TinMan is a highly productive and successful requisite step in deploying autonomous intelligence. TinMan AI Builder provides extremely fast and effective training of neural networks based on highly optimized algorithms. After a TinMan state model is added to the AI system, and given its desired inputs and outputs, it is ready to be trained. Training involves the presentment of a set of anticipated scenarios along with their respective desired outputs to the model. TinMan conveniently and automatically produces possible input sets. Learning begins when the “Learn Lessons” button is selected. Learning will usually be complete in seconds, and the model is ready (has knowledge). This knowledge is retained through the mathematical adjustment of the relationships among neurons and their connections within the model to produce all desired outcomes across all scenarios. Integrated scenarios spreadsheets provide significant enhancement to productivity by allowing "group-drag, select, and assign" output operations to a model scenarios list.
- Simulate - Test the AI System through Simulation
Simulation in TinMan AI Builder is possible once the AI system is constructed and the models have been adequately trained. Simulation is an important step in ensuring the AI system is performing the way it is designed to, and all models are responding to their respective scenarios as instructed. Simulation is accomplished by execution of the AI System Execute function on a desired frequency – typically anywhere from 5 -30 times per second. During simulation, the AI System visually displays status of variables, status of neural models, and currently selected outputs. The effectiveness of simulation is significantly enhanced through the use of several IDE simulation features, such as integrated action-specific sound, dynamic variables, integrated input controllers, live-wire view of active connections, breakpoints and others.
- Package - Produce the Runtime AI System
When you have finished designing, training and simulating the AI System, the system must now be put in its final form for runtime access in the host application. This process is referred to as Packaging. When packaging a system, the TinMan Packaging Wizard will step through the process, and ensure that all models throughout the new AI system are connected and trained. TinMan will also strip out all unnecessary elements, such as project items, visual elements and lengthy text descriptors and leaves just the multi-layer neural network and its described interfaces to inputs and outputs. Thus, size is reduced and speed is enhanced in the process. This file is now capable of being read by the runtime libraries, and thus utilized by the host application through simple class library functions for loading and executing.
Academic Proof Required
Win
Student 
School 
