An Improved Sampling Strategy for QBC Algorithm and its Application on Gas Sensor Array Signal Processing

This paper presents an improved sampling strategy for query by committee (QBC) algorithm based on inconsistency ranking for gas sensor array signal processing. For the stream-based query, the method measures the differences among the committee members' vote results of the samples from the unlabeled data pool and picks out only one sample with the largest degree of disagreement at once. For the pool-based query, the method ranks the query data according to their differences among the committee members' vote results and picks out a certain number of samples at the top at once. The experimental results showed that the approach only need very few initial training samples, and could obtain high accuracy after adding a small number of actively selected samples.