Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
6903383 | Applied Soft Computing | 2018 | 51 Pages |
Abstract
In this paper, we present a new method for evaluating the similarity of measured and predicted data which is of great importance to validate a prediction model. In particular, we provide a new similarity measure, which is used for the comparison of bidimensional arrays. The new similarity measure is employed in the context of the extremely low frequency magnetic field emitted by laptops. Accordingly, a model predicts the magnetic field levels emitted by the laptop. The experiment is performed on laptops working in normal operating condition and under stress. Measured and predicted levels are projected on the top and bottom surface of the laptops realizing bidimensional dangerousness maps. The similarity measure is used for the comparison of measured and predicted maps in normal operating condition and under stress, considering the differences in the levels and their disposition in the two maps. At the end, the similarity measure is employed for comparing different prediction models. The obtained results track a new direction in measurement and validation of magnetic field data.
Keywords
ELFACSMRMSIEEE62H30RMSDTCOSSD78A25RMSEMPCCPULSTMANNICNIRPSOMAlternating currentLong short-term memoryRoot mean square errorSimilarityArtificial Neural NetworkClassificationExtremely low frequencyLaptopSVMSupport vector machineroot mean square deviationroot mean squareMagnetic fieldHard diskcentral processing unit
Related Topics
Physical Sciences and Engineering
Computer Science
Computer Science Applications
Authors
Darko BrodiÄ, Alessia Amelio, Radmila JankoviÄ,