Abstract
Since off line hand-crafted conventional statistical feature selection/extraction methods are inefficient for signal-pattern classification of noisy sensor signals in manufacturing floor applications, we provide an automatic neural network approach to feature extraction, called Input Feature Scaling. Given only meaningful examples, the Input Feature Scaling algorithm enables a neural network, already trained by unsupervised competitive learning to cluster input patterns, to learn the relative importance of features for purposes of correct classification. The relative impor tance, expressed as weights, is adaptively learned in an additional supervised session. Experimental evaluation with both artificial data and real sensor data in tool-wear monitoring shows that our approach meets the requirements for "intelligent" sensor-signal processing in automated manufactur ing.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have