Models | EER(%) |
---|
 | Full feature set | Arm feature set |
---|
ML (16) | 9.1 | 15.9 |
MLLR (16) | 6.5 | 17.4 |
- The data presented are the average EER over four sessions for the LLR system with models created by ML and MLLR using left arm movement samples. A full feature set and an arm-only feature set were used to compare verification performance. The number within parentheses is the number of Gaussian mixtures in the GMM used for a given system configuration.