WORLD SCIENTIFIC eBooks
PATTERN CLASSIFICATION BASED ON QUANTUM NEURAL NETWORKS: A CASE STUDY
N.B. Karayiannis, R. Kretzschmar, Hans Richner
Abstract
This chapter presents the development, testing and evaluation of pattern classifiers designed using quantum neural networks (QNNs) to remove bird-contaminated wind profiler data. QNNs are trainable feedforward neural networks inherently capable of capturing and quantifying uncertainty in the training data. In this application, the pattern classifiers were developed by training QNNs to identify and remove bird-contaminated data recorded by a 1290-MHz wind profiler using a set of input features computed from the wind profiler measurements. A series of experiments were designed to evaluate several sets of features extracted from wind profiler data, compare various QNNs and traditional feedforward neural networks (FFNNs) of different sizes, and rate the criteria employed for identifying birds in wind profiler data based on the outputs of the trained neural networks. This experimental study indicates that QNN-based pattern classifiers can remove up to 90% of bird-contaminated wind profiler data, while QNNs are strong competitors to traditional FFNNs for real-world pattern classification applications.