Search

23 Feb 2006: However discriminative training1 has also been exam-ined within this framework [3, 4].

9 Dec 2006: For exam-ple, speaker models can be built for people who are likely to be inthe broadcast (such as prominent politicians or main news anchorsand reporters). ... However, this method relies on previously seen exam-ples of the test speakers, which is not

9 Dec 2006: For exam-ple, speaker models can be built for people who are likely to be inthe broadcast (such as prominent politicians or main news anchorsand reporters). ... However, this method relies on previously seen exam-ples of the test speakers, which is not

3 Apr 2006: Second, there are local spatial interac-tions between parts that can help with detection; for exam-ple, we expect to find the nose just above the mouth on aface.

22 Nov 2006: For exam-ple, the combination of the S2 and S3 GAUSS systems gave a 0.3%and 0.2% absolute improvements oneval03m and eval04 re-spectively.

23 Feb 2006: For exam-ple, the combination of the S2 and S3 GAUSS systems gave a 0.3%and 0.2% absolute improvements oneval03m and eval04 re-spectively.

14 Sep 2006: The results we show later, and the exam-ples in figure 3 confirm that there is indeed a practical advantage, but it is instructive toconsider on an intuitive level what the

22 Nov 2006: 14. 5 Multiple Stream Systems. One standard distributed representation, closely related to the scheme exam-ined here, is multiple stream modeling (20).

19 Mar 2006: W1 was found to be very large). Exam-ples are shown in Fig.

22 Nov 2006: The output of thismodel is a one-dimensional log-likelihood. To capture thedifferences in generative process between different exam-ples, each example is then mapped to the log-likelihoodgradient-space. ... Consider, for exam-ple, the second derivatives