Non-linear dimensionality reduction approaches, especially with AI, neural network techniques such as by [20, 63 65, 76], retain more of the relevant information and can improve classification accuracy at the same time.
It would thus be beneficial to incorporate this information into the formula, eliminating negative candidate events even before classification, while at the same time reducing the dimensionality of the datasets.
Experimental results based on benchmark and real-world data show that, compared to their competitors, our feature weighting approaches show higher classification accuracy, yet at the same time maintain the simplicity and lower execution time of the final models.
This is the hyperplane that separates the two classes in the space of descriptors, minimising the classification error and, at the same time, maximising the margin (i.e. the distances from the hyperplane to the closest samples of both classes, called support vectors): the idea is to find out the best trade off between accuracy and generalisation.
Embedded methods have the advantage that they include the interaction with the classification model, while at the same time being far less computationally intensive than wrapper methods.
This approach ensures no information loss, since the datasets are stored according to their original definitions and classification schemes, but at the same time enables data comparison.
The gains are two-fold: the classification accuracy increases and, at the same time, smaller networks can be used, reducing the required training and testing time.
The prediction of fault classification has been attempted at the same angular speed as the measured data as well as innovatively at the intermediate and extrapolated angular speed conditions, since it is not feasible to have measurement of vibration data at continuous speeds of interest.
