Antonette M. Logar

The use of neural network and discrete Fourier transform for real-time evaluation of friction stir welding

This paper introduces a novel real-time approach to detecting wormhole defects in friction stir welding in a nondestructive manner. The approach is to evaluate feedback forces provided by the welding process using the discrete Fourier transform and a multilayer neural network. It is asserted here that the oscillations of the feedback forces are related to the dynamics of the plasticized material flow, so that the frequency spectra of the feedback forces can be used for detecting wormhole defects. A one-hidden-layer neural network trained with the backpropagation algorithm is used for classifying the frequency patterns of the feedback forces. The neural network is trained and optimized with a data set of forge-load control welds, and the generality is tested with novel data set of position control welds. Overall, about 95% classification accuracy is achieved with no bad welds classified as good. Accordingly, the present paper demonstrates an approach for providing important feedback information about weld quality in real-time to a control system for friction stir welding.

An iterative method for training multilayer networks with threshold functions

Concerns the problem of finding weights for feed-forward networks in which threshold functions replace the more common logistic node output function. The advantage of such weights is that the complexity of the hardware implementation of such networks is greatly reduced. If the task to be learned does not change over time, it may be sufficient to find the correct weights for a threshold function network off-line and to transfer these weights to the hardware implementation. This paper provides a mathematical foundation for training a network with standard logistic function nodes and gradually altering the function to allow a mapping to a threshold unit network. The procedure is analogous to taking the limit of the logistic function as the gain parameter goes to infinity. It is demonstrated that, if the error in a trained network is small, a small change in the gain parameter will cause a small change in the network error. The result is that a network that must be implemented with threshold functions can first be trained using a traditional back propagation network using gradient descent, and further trained with progressively steeper logistic functions. In theory, this process could require many repetitions. In simulations, however, the weights have be successfully mapped to a true threshold network after a modest number of slope changes. It is important to emphasize that this method is only applicable to situations for which off-line learning is appropriate.

A comparison of recurrent neural network learning algorithms

Selected recurrent network training algorithms are described, and their performances are compared with respect to speed and accuracy for a given problem. Detailed complexity analyses are presented to allow more accurate comparison between training algorithms for networks with few nodes. Network performance for predicting the Mackey-Glass equation is reported for each of the recurrent networks, as well as for a backpropagation network. Using networks of comparable size, the recurrent networks produce significantly better prediction accuracy. The accuracy of the backpropagation network is improved by increasing the size of the network, but the recurrent networks continue to produce better results for the large prediction distances. Of the recurrent networks considered, Pearlmutters off-line training algorithm produces the best results.<<ETX>>

The ASTER polar cloud mask

This research is concerned with the problem of producing polar cloud masks for satellite imagery. The results presented are for Thematic Mapper (TM) data from the northern and southern polar regions, however, the techniques discussed will be applied to ASTER data when it becomes available. A series of classification techniques have been implemented and tested, the most promising of which is a neural network classifier. To use a neural network classifier, the pixels in the data must be transformed into feature vectors, some of which are used for training the network and the remainder of which are reserved for testing the final system. The first challenge is the identification of pure pixel samples from the imagery. The Interactive Visual Image Classification System (IVICS) was developed specifically for this project to make this task simpler for the human expert. After labeling the pixels, the feature vectors are generated. One hundred and forty potential vector elements, consisting of linear and nonlinear combinations of the satellite channel data, have been identified. Because smaller input vectors reduce the difficulty of training and can improve classification accuracy, the set of potential vector elements must be reduced. Two techniques have been tested: a histogram-based selection method and a fuzzy logic method. Both have proven effective for this task. Although the polar region is the only area considered in this work, a system that can produce cloud masks for all areas of the globe will be required. Thus, speed, extensibility, and flexibility requirements must be added to the accuracy constraint. To achieve these goals, a two-stage classification approach is used. The first stage uses a series of static and adaptive thresholds derived from statistical analysis of the polar scenes to reduce the set of possible classes to which a pixel may be assigned, once a cluster of classes has been selected, a neural network trained to distinguish between the classes in the cluster is used to make the ultimate classification.

Infectious disease and climate change: detecting contributing factors and predicting future outbreaks

The spread of infectious diseases worldwide is a cause for concern in areas traditionally susceptible to these diseases and in areas where these diseases have been previously unknown. This work concentrates on diseases for which global infection rates have been increasing and which are transmitted by mobile agents, or vectors. For example, the mosquito is the vector responsible for the transmission of malaria, dengue and viral encephalitis. Identification of the factors, particularly environmental factors which can be detected from satellite imagery, which are highly correlated to outbreaks of these diseases is an important aspect of this research. Development of a system which will monitor these factors, as well as short term climate variations, such as El Nino events, is also necessary to provide risk assessments for susceptible regions so that intervention strategies may be employed to prevent or limit the impact of many vector-borne diseases. This paper describes a geographical information system that has been populated with climatic, geographic, and disease data used to distinguish correlations between the different data sets. The system provides a graphical user interface that allows for a spatial representation of the number of disease cases to be overlaid on a variety of satellite-derived parameters and geographic data. The geographical information system is the tool which, when combined with satellite-derived products, provides a framework for studying disease outbreaks.

Embedding coupled oscillators into a feedforward architecture for improved time series prediction

The network defined by Hayashi (1994), like many purely recurrent networks, has proven very difficult to train to arbitrary time series. Many recurrent architectures are best suited for producing specific cyclic behaviors. As a result, a hybrid network has been developed to allow for training to more general sequences. The network used here is a combination of standard feedforward nodes and Hayashi oscillator pairs. A learning rule, developed using a discrete mathematics approach, is presented for the hybrid network. Significant improvements in prediction accuracy were produced compared to a pure Hayashi network and a backpropagation network. Data sets used for testing the effectiveness of this approach include Mackey-Glass, sunspot, and ECG data. The hybrid models reduced training and testing error in each case by a least 34%.

An extension to the Hayashi coupled oscillator network training rule

A variety of recurrent network architectures have been developed and applied to the problem of time series prediction. One particularly interesting network was developed by Hayashi (1994). Hayashi presented a network of coupled oscillators and a training rule for the network. His derivation was based on continuous mathematics and provided a mechanism for updating the weights into the output nodes. The work presented here gives an alternative derivation of Hayashis learning rule based on discrete mathematics as well an extension to the learning rule which allows for updating of all weights in the network.