Search

3 Sep 2010: In practice this is rarely the case. Originally the backgroundnoise was simply estimated from periods of “silence” in the test conditions. ... In particular the AURORA2 test set [27] has been used to evaluate a number of systems within a

25 Oct 2010: The mixture model makes predictionsfor arbitrary new test points and typically has a relativelysmall number of parameters. ... where NG is the number ofground truth images in the test data set).

25 Oct 2010: An inde-pendent illumination set comprising both training and test sets. Fig. ... Fig. 6 compares the recognition accuracy ofall methods, where the experiment numbers correspond to thecombinations of the training/test lighting sets.

25 Oct 2010: An inde-pendent illumination set comprising both training and test sets. Fig. ... Fig. 6 compares the recognition accuracy ofall methods, where the experiment numbers correspond to thecombinations of the training/test lighting sets.

12 Jan 2010: diversion, to test the claim in Section 2.2 that constraining y1 is the best strategy for condi-tioning the model-fitting process.