Integrating Neural Networks and Semi-Markov Processes for Automated Knowledge Acquisition: An Application to Real-time Scheduling* zy Ting-Peng Liang and Herbert Moskowitz zyxwv Krannert School of Management, Purdue University, West Lafayette, IN 47907 Yuehwern Yih School of Industrial Engineering, Purdue University, West Lafayette, IN 47907 ABSTRACT Recently, artificial neural networks (ANN) have gained attention as a promising modeling tool for building intelligent systems. A number of applications have been reported in areas varying from pattern recognition to bankruptcy prediction. In this paper, we present a creative methodology that integrates computer simulation, semi-Markov optimization, and ANN techniques for auto- mated knowledge acquisition in real-time scheduling. The integrated approach focuses on the synergy between operations research and ANN in eliciting human knowledge, filtering inconsistent data, and building competent models capable of performing at the expert level. The new approach includes three main components. First, computer simulation is used to collect expert decisions. This step allows expert knowledge to be obtained in a non-intrusive way and minimizes the difficulties involved in interviewing experts, constructing repertory grids, or using other similar structures required for manual knowledge acquisition. The data collected from computer simulation are then optimized using a semi-Markov decision model to remove data redundancies, inconsis- tencies, and errors. Finally, the optimized data are used to build ANN-based expert systems. The integrated approach is evaluated by comparing it with the human expert and using ANN alone in the domain of real-time scheduling. The results indicate that ANN-based systems perform wone than human experts from whom the data were collected, but the integrated approach outperforms human experts and ANN models alone. zyxwvu Subject Areas: Decision Support zyxwvut Systems, Learning Modeis, Markov Processes, and Scheduling. INTRODUCTION Knowledge acquisition is a well-known bottleneck in building expert systems. Due to human cognitive limitations, experts often have difficulties in articulating their knowledge. Recently, much research has focused on using machine learning methods that induce knowledge directly from actual expert decisions to alleviate the problem. Applications have been reported in many domains (e.g., [l], [2], zyx [14], [18], [22], and [26]). When reliable induction methods are available, the machine learning approach can simplify the knowledge acquisition process to two major tasks: collecting expert decisions (called training data) and inducing knowledge from the collected data. *This research was supported by grants from AT&T Foundation, Purdue Research Foundation and the Krannert School’s Center for the Management of Manufacturing Enterprises. The project is a team effort but the simulation and semi-Markov analysis software are primarily due to Yuehweni Yih. The authors also thank Heather Perea and Hari Sankar for their research assistance. 1297