| Autor: |
Hall, Richard, Rathod, Hemang, Maiorca, Mauro, Ioannou, Ioanna, Kazmierczak, Edmund, O΄Leary, Stephen, Harris, Peter |
| Zdroj: |
Haptic & Audio Interaction Design (9783540878827); 2008, p50-59, 10p |
| Abstrakt: |
It is desirable to automatically classify data samples for the assessment of quantitative performance of users of haptic devices as the haptic data volume may be much higher than is feasible to manually annotate. In this paper we compare the use of three k-metrics for automated classifaction of human motion: cosine, extrinsic curvature and symmetric centroid deviation. Such classification algorithms make predictions about data attributes, whose quality we assess via three mathematical methods of comparison: root mean square deviation, sensitivity error and entropy correlation coefficient. Our assessment suggests that k-cosine might be more promising at analysing haptic motion than our two other metrics. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
