Detection error tradeoff

A detection error tradeoff (DET) graph is a graphical plot of error rates for binary classification systems, plotting false reject rate vs. false accept rate. The x- and y-axes are scaled non-linearly by their normal deviates, yielding tradeoff curves that are more linear than ROC curves, and spend most of the image area highlighting the differences of importance in the critical operating region.

Axis warping

The normal deviate mapping (or normal quantile function, or inverse normal cumulative distribution) is given by the probit function, so that the horizontal axis is x = probit(P_fa) and the vertical is y = probit(P_fr), where P_fa and P_fr are the false-accept and false-reject rates.

The probit mapping maps probilities which live in the unit interval [0,1], to the extended real line [−∞, +∞]. Since this makes the axes infinitely long, one has to confine the plot to some finite rectangle of interest.

External links

• Example DET graph
• Many example DET graphs (at bottom)

See also

• Receiver operating characteristic
• Constant false alarm rate
• Detection theory
• Expected performance curve
• False alarm