A New Optimization Approach to Resource Distribution Using Semi-Supervised Learning Graphs

An effective Post-Disaster Management System (PDMS) will result in distribution of emergency resources such as, hospital, storage and transportation in a reasonable and equitable manner. This study starts with semi-supervised learning (SSL) based graph system to provide post-disaster path optimizations. Next, the graph-based resource is converted to a directed graph resulting in an adjacent matrix. Decision information is provided in two ways: clustering operation and graph-based semi-supervised optimization. The PDMS in this study incorporates a path optimization algorithm based on Ant Colony Optimization (ACO) that results in costeffective resource distribution. Simulation results demonstrate the effectiveness of the proposed methodology by comparing ACO with clustering based algorithms of tour improvement algorithm (TIA) and Min-Max Ant System (MMAS).

[1]  Bernhard Schölkopf,et al.  Introduction to Semi-Supervised Learning , 2006, Semi-Supervised Learning.

[2]  Zoubin Ghahramani,et al.  Combining active learning and semi-supervised learning using Gaussian fields and harmonic functions , 2003, ICML 2003.

[3]  Jiqiang Chen,et al.  A Special Supervised Learning Algorithm and Its Applications , 2009, 2009 Ninth International Conference on Hybrid Intelligent Systems.

[4]  Thomas Stützle,et al.  MAX-MIN Ant System , 2000, Future Gener. Comput. Syst..

[5]  Kefei Zhang,et al.  An object-relational prototype of GIS-based disaster database , 2009 .

[6]  Shih-Fu Chang,et al.  Graph transduction via alternating minimization , 2008, ICML '08.

[7]  Tian Zhao,et al.  Automatic search of geospatial features for disaster and emergency management , 2010, Int. J. Appl. Earth Obs. Geoinformation.

[8]  H. J. Scudder,et al.  Probability of error of some adaptive pattern-recognition machines , 1965, IEEE Trans. Inf. Theory.

[9]  Rey-Chue Hwang,et al.  A Hybrid Supervised Neural Network Learning Algorithm , 2008, 2008 3rd International Conference on Innovative Computing Information and Control.