ISCA Archive Interspeech 2013
ISCA Archive Interspeech 2013

Likelihood-ratio calibration using prior-weighted proper scoring rules

Niko Brümmer, George R. Doddington

Prior-weighted logistic regression has become a standard tool for calibration in speaker recognition. Logistic regression is the optimization of the expected value of the logarithmic scoring rule. We generalize this via a parametric family of proper scoring rules. Our theoretical analysis shows how different members of this family induce different relative weightings over a spectrum of applications of which the decision thresholds range from low to high. Special attention is given to the interaction between prior weighting and proper scoring rule parameters. Experiments on NIST SREf12 suggest that for applications with low false-alarm rate requirements, scoring rules tailored to emphasize higher score thresholds may give better accuracy than logistic regression.


doi: 10.21437/Interspeech.2013-470

Cite as: Brümmer, N., Doddington, G.R. (2013) Likelihood-ratio calibration using prior-weighted proper scoring rules. Proc. Interspeech 2013, 1976-1980, doi: 10.21437/Interspeech.2013-470

@inproceedings{brummer13_interspeech,
  author={Niko Brümmer and George R. Doddington},
  title={{Likelihood-ratio calibration using prior-weighted proper scoring rules}},
  year=2013,
  booktitle={Proc. Interspeech 2013},
  pages={1976--1980},
  doi={10.21437/Interspeech.2013-470}
}