Abstracts - faqs.org

Abstracts

Business, international

Search abstracts:
Abstracts » Business, international

A neural network model for scheduling problems

Article Abstract:

An artificial neural network architecture for solving the NP-hard scheduling problems was developed. The architecture, which employs an external processor to 'monitor and control the evolution of the network,' was applied to the single-machine mean tardiness scheduling problem and the minimum makespan job shop scheduling problem. According to test runs, the new procedure is superior to the Wilkerson-Irwin algorithm for the single-machine problem. In addition, it converges to an optimal solution for most job shop scheduling problems.

Author: Sabuncuoglu, Ihsan, Gurgun, Burchkaan
Publisher: Elsevier B.V.
Publication Name: European Journal of Operational Research
Subject: Business, international
ISSN: 0377-2217
Year: 1996
Scheduling (Management)

User Contributions:

Comment about this article or add new information about this topic:

CAPTCHA


A comparison of neural networks and linear scoring models in the credit union environment

Article Abstract:

The performance of feedforward neural networks and traditional statistics-based discriminant analysis methods in constructing credit scoring models using the database of three US credit unions was compared. Specifically, the predictive ability of multilayer perceptrons and modular neural networks was evaluated. The results indicated that customized neural networks are superior only when the performance criteria are a percentage of bad loans correctly classified.

Author: Desai, Vijay S., Crook, Jonathan N., Overstreet, George A., Jr.
Publisher: Elsevier B.V.
Publication Name: European Journal of Operational Research
Subject: Business, international
ISSN: 0377-2217
Year: 1996
Models, Credit ratings, Discriminant analysis

User Contributions:

Comment about this article or add new information about this topic:

CAPTCHA


A new neural network for solving linear programming problems

Article Abstract:

A simple and low-cost analog neural network architecture for solving linear programming (LP) problems is described. The universal and flexible network architecture, which may be easily implemented using very-large-scale-integration technology, is based on nonlinear differential equation representations of LP problems. Its main components are a single neuron, adaptive synaptic weights which are modified by a learning algorithm and a pre-processing circuit.

Author: Cichocki, A., Unbehauen, R., Weinzierl, K., Holzel, R.
Publisher: Elsevier B.V.
Publication Name: European Journal of Operational Research
Subject: Business, international
ISSN: 0377-2217
Year: 1996
Linear programming, Parallel processing

User Contributions:

Comment about this article or add new information about this topic:

CAPTCHA


Subjects list: Operations research, Research, Management science, Computer networks, Case studies, Neural networks
Similar abstracts:
  • Abstracts: On genetic algorithms for the packing of polygons. Regular packing of congruent polygons on the rectangular sheet
  • Abstracts: Giant sucking sound? Concerns about hollowing out in South Korea. Out of patience: South Korean government intervenes in Kia crisis
  • Abstracts: Israeli sales continue despite delays for El Al. Air France minority stake to float. New privatisation plan for Polish banks approved
  • Abstracts: Hong Kong may be losing competitive edge to Singapore. Australia's competitive edge.: economy makes strides and ranks high on many international measures
  • Abstracts: Don't give in to high yields. Currency and stock markets are climbing out of a funk
This website is not affiliated with document authors or copyright owners. This page is provided for informational purposes only. Unintentional errors are possible.
Some parts © 2025 Advameg, Inc.