Title :
Inhibitory grids and the assignment problem
Author :
Wolfe, William J. ; MacMillan, James M. ; Brady, George ; Mathews, Robert ; Rothman, Jay Alan ; Mathis, Donald ; Orosz, Michael Donald ; Anderson, Charlie ; Alaghban, Gila
Author_Institution :
Dept. of Comput. Sci. & Eng., Colorado Univ., Denver, CO, USA
fDate :
3/1/1993 12:00:00 AM
Abstract :
A family of symmetric neural networks that solve a simple version of the assignment problem (AP) is analyzed. The authors analyze the suboptimal performance of these networks and compare the results to optimal answers obtained by linear programming techniques. They then use the interactive activation model to define the network dynamics-a model that is closely related to the Hopfield-Tank model. A systematic analysis of hypercube corner stability and eigenspaces of the connection strength matrix leads to network parameters that give feasible solutions 100% of the time and to a projection algorithm that significantly improves performance. Two formulations of the problem are discussed: (i) nearest corner: encode the assignment numbers as initial activations, and (ii) lowest energy corner: encode the assignment numbers as external inputs
Keywords :
eigenvalues and eigenfunctions; matrix algebra; neural nets; operations research; Hopfield-Tank model; assignment problem; connection strength matrix; eigenspaces; hypercube corner stability; interactive activation model; network dynamics; operations research; symmetric neural networks; Algorithm design and analysis; Analog circuits; Computer science; Hypercubes; Linear programming; Neural networks; Performance analysis; Projection algorithms; Random number generation; Stability analysis;
Journal_Title :
Neural Networks, IEEE Transactions on
