Essay: Hierarchical knowledge structure

‘Hierarchical knowledge structure’ should be build between AI decision system and physical device. The concept of a hierarchical knowledge structure is to decompose a complicated decision into series of smaller decisions of increasing level of details. In other words, many small separate decisions can contribute to solve a complex over all decision. The operation of the hierarchical knowledge structure is such that it makes a decision in a top-down approach, moving from general to specific.
Burns, [b] introduces another system structure of intelligent machine. An intelligent machine may be regarded as a system that comprises of a number of essential subsystems:
1. The perception subsystem collects information about the environment, and the system itself. It then processes the information to provide data on the current state of the machine, and the world in which it operates. The essential elements are:
Sensor array: provides raw data on the environment and the machine.
Signal processing: transforms the information into a suitable form.
2. The cognition subsystem is concerned with making decisions on actions that the machine should undertake to achieve specified goals. This decision-making process takes place usually under conditions of uncertainty. In order to do this the machine must be able to model itself within its environment and be able to predict results of its own actions, and also how the environment may change.
3. Fuzzy logic system was first proposed by Zadeh in 1965 and is based on the concept of fuzzy sets. A fuzzy set is a set whose membership function takes values between zero and one. The range of fuzzy sets in a given window is called the “universe of discourse”. One advantage of fuzzy logic control is that robust control can be achieved for a system whose dynamic characteristics are unknown. Unlike conventional linear logic, it is possible for a parameter to have a membership of more than one set.
4. Neural network systems
An artificial neural network is designed to learn in the same way as a human brain.
A psychology experiment conducted by Professor Naomi Ehrich Leonard [c] found a convergence of the aggregate human decision-making behavior to reward structures with matching points. In this experiment, the human subject in the psychology experiments chooses between two options at regular time intervals and receives a reward after each choice that depends on recent past decisions. From the diagram, human tend to adopt strategies that bring them close to the matching point. However, the matching point is not the optimal reward. By analyzing the results of human psychology experiment, AI automation systems can be built to make optimal decisions.