|
1. |
PREDICTING BANK FAILURES: A NEURAL NETWORK APPROACH |
|
Applied Artificial Intelligence,
Volume 4,
Issue 4,
1990,
Page 265-282
KARYAN TAM,
MELODY KIANG,
Preview
|
PDF (549KB)
|
|
摘要:
The purpose of this paper is to present a neural network approach to predicting bank failures and to compare it with existing prediction methods. The task of constructing a prediction model is cast as one of training a network with a set of bankruptcy cases. Empirical results show that neural network is a competitive method among existing ones in assessing the likelihood of bank failures, especially in reducing type I misclassification rate. Issues relating to the potential and limitations of.neural network as a modeling tool are also addressed.
ISSN:0883-9514
DOI:10.1080/08839519008927951
出版商:Taylor & Francis Group
年代:1990
数据来源: Taylor
|
2. |
BUILDING TOOLS FOR SOFTWARE ENGINEERING WITH ARTIFICIAL INTELLIGENCE TECHNIQUES |
|
Applied Artificial Intelligence,
Volume 4,
Issue 4,
1990,
Page 283-307
DANIELE NARDI,
MARCO TUCCI,
Preview
|
PDF (791KB)
|
|
摘要:
In this paper we address the problem of building tools for improving the software engineering process by taking advantage of artificial intelligence techniques. More specifically, we provide a representation of the evolution of a software project by means of temporal logics. Such a representation is the basis for the construction of tools for the control and management of a software project. A system for Access and Version Control (SCAV) provides a suitable test bed for our approach. In the paper we present the formalization of the history of a software project developed under SCAV, and we show how several kinds of functionalities can be easily realized. The formalization is built within the framework of the event calculus in such a way that every step of the project development determined by the execution of a SCAV operation is described as an event in the calculus. We present an implementation of the system in PROLOG, which allows for a direct and natural formulation of the event calculus rules.
ISSN:0883-9514
DOI:10.1080/08839519008927952
出版商:Taylor & Francis Group
年代:1990
数据来源: Taylor
|
3. |
ANALYSIS AND REPRESENTATION OF NEUROANATOMICAL KNOWLEDGE |
|
Applied Artificial Intelligence,
Volume 4,
Issue 4,
1990,
Page 309-336
JÖRG NIGGEMANN,
Preview
|
PDF (752KB)
|
|
摘要:
ANATOM contains anatomical knowledge, using three different representation formalisms. The first is propositional representation (frames, constraints, logic), the second is a two-dimensional depictional representation (cell matrix), and the third is a three-dimensional model. Each has its own module for knowledge acquisition. The analysis of anatomical knowledge leads to meta-knowledge about the structure of anatomical knowledge. This is used to construct the knowledge acquisition tool Anatomy Description Language (ADL). Special attention is paid to the representation of the variations of the vessel system. This paper discusses the analysis of the anatomical knowledge and the propositional knowledge base and introduces briefly the concepts of depictional representation of neural fiber connections (without discussing depictional representation in general) and describes some problems of spatial inferences on three-dimensional anatomical structure.
ISSN:0883-9514
DOI:10.1080/08839519008927953
出版商:Taylor & Francis Group
年代:1990
数据来源: Taylor
|
4. |
REASONING FOR AUTOMATED MODEL INTEGRATION |
|
Applied Artificial Intelligence,
Volume 4,
Issue 4,
1990,
Page 337-358
TING-PENG LIANG,
Preview
|
PDF (522KB)
|
|
摘要:
This paper investigates some reasoning issues involved in developing an integrated modeling environment called a model management system. A model management system is a computer-based environment designed to support effective development and utilization of quantitative decision models. Since the construction of decision models is a knowledge-intensive process, reasoning plays a critical role. Reasoning is particularly important when automated model integration is needed in a large-scale system. In this case, heuristics are required to reduce the complexity of the process. This paper examines the planning and automated formulation of complex models from smaller building blocks. First, a hierarchy of abstractions that integrates previous contributions in model management is presented. Then, a modeling process is formulated as a planning process by which a set of operators are scheduled to achieve a specific goal. This process involves searches for alternatives that can eliminate the difference between the initial stale and the goal state. Various reasoning strategies and heuristic evaluation Junctions that can be used to improve the efficiency of developing a master plan for model integration are discussed.
ISSN:0883-9514
DOI:10.1080/08839519008927954
出版商:Taylor & Francis Group
年代:1990
数据来源: Taylor
|
|