Role Of Artificial Neural Networks Research Articles

Artificial Neural Networks in Hydrology. I: Preliminary Concepts

In this two-part series, the writers investigate the role of artificial neural networks (ANNs) in hydrology. ANNs are gaining popularity, as is evidenced by the increasing number of papers on this ...

Artificial neural networks in bankruptcy prediction: General framework and cross-validation analysis

In this paper, we present a general framework for understanding the role of artificial neural networks (ANNs) in bankruptcy prediction. We give a comprehensive review of neural network applications in this area and illustrate the link between neural networks and traditional Bayesian classification theory. The method of cross-validation is used to examine the between-sample variation of neural networks for bankruptcy prediction. Based on a matched sample of 220 firms, our findings indicate that neural networks are significantly better than logistic regression models in prediction as well as classification rate estimation. In addition, neural networks are robust to sampling variations in overall classification performance.

PREDICTING WATER QUALITY IN DISTRIBUTION SYSTEMS USING ARTIFICIAL NEURAL NETWORKS.

The possible role of artificial neural networks to model the impact of the distribution network on water quality under different patterns of demand and for different source water qualities is considered and contrasted against a deterministic approach. To investigate the feasibility of this technique, a case study is presented of the prediction of oxidation reduction potential at a single point and at multiple points within a distribution network using hydraulic and water quality data collected at the inlet to the network. Based on the results of the study, the advantages and disadvantages of an artificial neural network approach and a deterministic approach to the prediction of water quality in distribution are identified. The study has demonstrated the viability of the application of artificial neural networks to predict water quality changes in distribution, and it has been shown that water quality changes within the distribution network are essentially flow-driven. The potential of this technique to evaluate the relative importance of parameters prior to the formulation of a deterministic model was highlighted. The need to complete further work to ascertain the extent of data required to achieve a given accuracy of prediction by an artificial neural network in this particular role has also been identified.

Recognition of daily life motor activity classes using an artificial neural network

Objective: To investigate a possible role of artificial neural networks for the automated recognition and classification of daily life activities (eg, sitting, lying, standing, walking, etc) in an attempt to reduce the cost of manual recognition and classification. Methods: Data from sessions of about 10 hours of continuous recording of eight ambulatory patients were used to train and evaluate eight probabilistic neural networks, each of which is configured for one subject. To provide the reference data for building the training set, the instrumented subject follows a 15-to 30-minute protocol consisting of several daily life activities. To properly evaluate the networks, the remaining manually labeled data of each subject were compared with the output of each trained network. Results: The average recognition rate of the trained neural networks was equal to 95% good classification of all presented cases of the daily life activity. Automatic misclassification of 5% resulted from certain activities being too short or the occurrence of activities that were not included in the training set. Conclusion: The preliminary results of the trained neural networks have indicated that the probabilistic neural network is a potentially useful tool for the recognition of daily life motor activities.

Artificial neural networks in power systems. Part 2: Types of artificial neural networks

This tutorial describes the different types of artificial neural networks in use today. It is the second in a series of three tutorials which describe the role of artificial neural networks in solving complex problems in modern-day power systems.

A feedforward neural network for multiple criteria decision making

Many complex real-world problems are characterized as decision making with multiple, conflicting and noncommensurate objectives. Because of the complexity of factors that are involved, it is usually difficult to derive a decision rule for determining the most desirable alternative. This paper is to demonstrate the potential role of artificial neural networks for multiple criteria decision making. This paper presents a feedforward neural network for solving discrete multiple criteria decision problems under certainty. Starting with formulating multiple criteria decision problems under the theme of supervised learning, this paper specifies two types of multiattribute decision models, proposes a particular form of feedforward neural network, analyzes some desirable properties associated with supervised learning, presents an improved learning algorithm and discusses results of illustrative examples and numerical simulation.

Preprocessing inputs for adaptive critic control

Preliminary results reported in this paper constitute the first step in an effort to explore the role of artificial neural networks (ANN) in maintaining the stability of dynamic systems. In particular, the goal is to understand the control mechanism necessary to maintain the postural stability of a musculoskeletal model of a human. In view of the speculation that the olive nucleus in the brainstem acts as an adaptive critic for the descending motor control, we chose to investigate the adaptive critic algorithm. The adaptive critic is a reinforcement learning technique which utilizes qualitative feedback, instead of quantitative feedback as in supervised learning; or no feedback at all as in self-organizing systems. The use of the adaptive critic to control a dynamic system, such as a cart-pole system, has been demonstrated by a number of investigators including Widrow, Gupta, and Maitra (1973); Barto, Sutton, and Anderson (1983); and Anderson (1989). Barto quantized the state variables before processing them in the adaptive critic network. Anderson's approach essentially replaces the quantizer with an ANN. In this paper, Anderson's method is modified by incorporating into the control loop what Klassen and Pao refer to as the Functional Link Outerproduct (FLO) in place of Anderson's ANN. The FLO expands the original input space by including higher order terms. This stratagem, while simplifying the structure, also improves the learning rate of the network. While Anderson's algorithm, when implemented as a controller, generalizes its control much better than Barto's algorithm, it takes much longer to train. The FLO, on the other hand, provides improved generalization over Barto's algorithm without the large increase in training time. The long-term goal of this project is to explore the utility of these ideas in controlling human posture.