%Aigaion2 BibTeX export from Idiap Publications
%Saturday 21 December 2024 05:35:52 PM

@INPROCEEDINGS{Dey_ICASSP-2_2016,
         author = {Dey, Subhadeep and Madikeri, Srikanth and Motlicek, Petr},
       projects = {Idiap, SIIP},
          month = mar,
          title = {INFORMATION THEORETIC CLUSTERING FOR UNSUPERVISED DOMAIN-ADAPTATION},
      booktitle = {Proceedings of 2016 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2016)},
           year = {2016},
          pages = {5580-5584},
      publisher = {IEEE},
       location = {Shanghai},
       crossref = {Dey_Idiap-RR-09-2016},
       abstract = {The aim of the domain-adaptation task for speaker verification is
to exploit unlabelled target domain data by using the labelled source
domain data effectively. The i-vector based Probabilistic Linear Dis-
criminant Analysis (PLDA) framework approaches this task by clus-
tering the target domain data and using each cluster as a unique
speaker to estimate PLDA model parameters. These parameters are
then combined with the PLDA parameters from the source domain.
Typically, agglomerative clustering with cosine distance measure is
used. In tasks such as speaker diarization that also require unsuper-
vised clustering of speakers, information-theoretic clustering mea-
sures have been shown to be effective. In this paper, we employ
the Information Bottleneck (IB) clustering technique to find speaker
clusters in the target domain data. This is achieved by optimizing the
IB criterion that minimizes the information loss during the cluster-
ing process. The greedy optimization of the IB criterion involves ag-
glomerative clustering using the Jensen-Shannon divergence as the
distance metric. Our experiments in the domain-adaptation task in-
dicate that the proposed system outperforms the baseline by about
14\% relative in terms of equal error rate.},
            pdf = {https://publications.idiap.ch/attachments/papers/2016/Dey_ICASSP-2_2016.pdf}
}



crossreferenced publications: 
@TECHREPORT{Dey_Idiap-RR-09-2016,
         author = {Dey, Subhadeep and Madikeri, Srikanth and Motlicek, Petr},
       projects = {Idiap, SIIP},
          month = {4},
          title = {INFORMATION THEORETIC CLUSTERING FOR UNSUPERVISED DOMAIN-ADAPTATION},
           type = {Idiap-RR},
         number = {Idiap-RR-09-2016},
           year = {2016},
    institution = {Idiap},
       abstract = {The aim of the domain-adaptation task for speaker verification is
to exploit unlabelled target domain data by using the labelled source
domain data effectively. The i-vector based Probabilistic Linear Dis-
criminant Analysis (PLDA) framework approaches this task by clus-
tering the target domain data and using each cluster as a unique
speaker to estimate PLDA model parameters. These parameters are
then combined with the PLDA parameters from the source domain.
Typically, agglomerative clustering with cosine distance measure is
used. In tasks such as speaker diarization that also require unsuper-
vised clustering of speakers, information-theoretic clustering mea-
sures have been shown to be effective. In this paper, we employ
the Information Bottleneck (IB) clustering technique to find speaker
clusters in the target domain data. This is achieved by optimizing the
IB criterion that minimizes the information loss during the cluster-
ing process. The greedy optimization of the IB criterion involves ag-
glomerative clustering using the Jensen-Shannon divergence as the
distance metric. Our experiments in the domain-adaptation task in-
dicate that the proposed system outperforms the baseline by about
14\% relative in terms of equal error rate.},
            pdf = {https://publications.idiap.ch/attachments/reports/2016/Dey_Idiap-RR-09-2016.pdf}
}