BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Cluster Adaptive Training for Deep Neural Networks - Tian Tan (vis
 itor at CUED)
DTSTART:20141128T120000Z
DTEND:20141128T130000Z
UID:TALK56176@talks.cam.ac.uk
CONTACT:Rogier van Dalen
DESCRIPTION:Although context-dependent DNN-HMM systems have achieved signi
 ficant improvements over GMM-HMM systems\, there still exists big performa
 nce degradation if the acoustic condition of the test data mismatches that
  of the training data. Hence\, adaptation and adaptive training of DNN are
  of great research interest. Previous works mainly focus on adapting the p
 arameters of a single DNN. These methods all require relatively large numb
 er of parameters to be estimated during adaptation. In contrast\, this pap
 er employs the cluster adaptive training (CAT) framework for DNN adaptatio
 n. Here\, multiple DNNs are constructed to form the bases of a canonical p
 arametric space. During adaptation\, an interpolation vector\, specific to
  a particular acoustic condition\, is used to combine the multiple DNN bas
 es into a single adapted DNN. \n\n_Biography_\n\nTian Tan is a Ph.D studen
 t with Kai Yu at Shanghai Jiao Tong University (China). He is visiting the
  University of Cambridge till 5th March. \n\n*Sandwiches will be provided*
LOCATION:Department of Engineering - LR10
END:VEVENT
END:VCALENDAR