A Neural Network Approach to Entropy Maximizing Spatial Interaction Analysis

Principal Investigator:
Leung, Yee

Co-investigator(s):
: ---

Summary:
The objective of the proposed investigation is to formulate a neural-network approach to the efficient and effective analysis of large-scale spatial interaction problems based on the concept entropy maximization. The proposed framework will give direct and real-time solution to convex entropy-maximization problems with nice theoretical properties and high practical values. The study will pave the road for the construction of a more general neural network approach to solve convex, concave and global entropy-maximization models often encountered in real-life decision making. It will also lead to the development of a user-friendly software environment for the planning of origin-destination types of spatial interaction databases, solution algorithms, and graphic displays.