--- abstract: "The discrimination technique for estimating parameters of Gaussian mixtures that is based on the Extended Baum-Welch transformations (EBW) has had significant impact on the speech recognition community. \r\nIn this paper we introduce a general definition of a family of EBW transformations that can be associated with a weighted sum of updated and initial models. We compute a gradient steepness measurement for a family of EBW transformations that are applied to functions of Gaussian mixtures and demonstrate the growth property of these transformations. We consider EBW transformations of discriminative functions in which EBW controlled parameters are adapted to a gradient steepness measurement or to the likelihood of the data given the model. We present experimental results that show that adapted EBW transformations can significantly speed up estimating parameters of Gaussian mixtures and give better decoding results." altloc: [] chapter: ~ commentary: ~ commref: ~ confdates: ~ conference: ~ confloc: ~ contact_email: ~ creators_id: - kanevsky@us.ibm.com - dpovey@us.ibm.com - bhuvana@us.ibm.com - tsainath@MIT.EDU creators_name: - family: Kanevsky given: Dimitri honourific: Dr lineage: '' - family: 'Povey ' given: 'Daniel ' honourific: Dr lineage: '' - family: Ramabhadran given: 'Bhuvana ' honourific: Dr lineage: '' - family: Sainath given: Tara honourific: Dr lineage: '' date: 2007-10-01 date_type: completed datestamp: 2008-01-15 23:56:42 department: ~ dir: disk0/00/00/59/02 edit_lock_since: ~ edit_lock_until: ~ edit_lock_user: ~ editors_id: [] editors_name: [] eprint_status: archive eprintid: 5902 fileinfo: /style/images/fileicons/application_pdf.png;/5902/1/rc24458.pdf full_text_status: public importid: ~ institution: ~ isbn: ~ ispublished: ~ issn: ~ item_issues_comment: [] item_issues_count: 0 item_issues_description: [] item_issues_id: [] item_issues_reported_by: [] item_issues_resolved_by: [] item_issues_status: [] item_issues_timestamp: [] item_issues_type: [] keywords: 'MMIE training, EBW transformations, gradient steepness' lastmod: 2011-03-11 08:57:02 latitude: ~ longitude: ~ metadata_visibility: show note: ~ number: ~ pagerange: ~ pubdom: TRUE publication: ~ publisher: ~ refereed: FALSE referencetext: "S. Axelrod, V. Goel, R. Gopinath, P. Olsen, and K. Visweswariah, \"Discriminative Training of Subspace Constrained GMMs for Speech Recognition,\" to be submitted to IEEE Transactions on Speech and Audio Processing. \r\n\r\nL.E.Baum and J.A. Eagon, \"An inequality with applications to statistical prediction for functions of Markov processes and to a model of ecology,\" {\\em Bull. Amer. Math. Soc.}, vol. 73, pp.360-363, 1967.\r\n\r\nA. Gunawardana and W. Byrne, ``Discriminative Speaker Adaptation with Conditional Maximum Likelihood Linear Regression,'' ICASSP, 2002.\r\n\r\nP.S. Gopalakrishnan, D. Kanevsky, D. Nahamoo and A. Nadas, \"An inequality for rational functions with applications to some statistical estimation problems\", IEEE Trans. Information Theory, Vol. 37, No.1 January 1991 .\r\n\r\nD. Kanevsky, \"Growth Transformations for General Functions\", RC22919 (W0309-163), September 25, 2003.\r\n\r\nD. Kanevsky, \"Extended Baum transformations for general functions\", in Proc. ICASSP, 2004.\r\n\r\nD. Kanevsky, ``Extended Baum Transformations for General Functions, II\", tech. Rep. RC23645(W0506-120), Human Language technologies, IBM , 2005, http://cogprints.org/5058/01/rc23645.pdf .\r\n\r\nD. Kanevsky, \"Constrained corrective training for continuous parameter system\", US patent 6,044,344, March 28, 2000. \r\n\r\nCong Liu Peng Liu Hui Jiang Soong, F. Ren-Hua Wang, \"A Constrained Line Search Optimization for Discriminative Training in Speech Recognition\",\r\nin Proc. ICASSP, 2007.\r\n\r\nY. Normandin, \"An improved MMIE Training Algorithm for Speaker Independent, Small Vocabulary, Continuous Speech Recognition\", Proc. ICASSP'91, pp. 537-540, 1991.\r\nDaniel Povey, \"Discriminative Training for Large Vocabulary Speech Recognition\", PhD Thesis March 1, 2003.\r\n\r\nDaniel Povey, Dimitri Kanevsky, Brian Kingsbury, Bhuvana Ramabhadran, George Saon and Karthik Visweswariah \"Boosted MMI for model and feature-space discriminative training\", submitted for ICASSP'08.\r\n\r\nTara N. Sainath, Dimitri Kanevsky, Giridharan Iyengar,``Unsupervised Audio Segmentation Using Extended Baum-Welch Transformations, in Proc. ICASSP, 2007.\r\nTara N. Sainath, Victor Zue, Dimitri Kanevsky, ``Audio-Classification using Extended Baum-Welch Transformations\" , in Proc. Interspeech 2007.\r\n\r\nTara N. Sainath, Dimitri Kanevsky, Bhuvana Ramabhadran ``Broad Phoentic Recognition in a Hidden Markov Model Framework Using Extended Baum-Welch Transformations \" , to appear in Proc. ASRU 2007.\r\n\r\nTara N. Sainath, Dimitri Kanevsky, Bhuvana Ramabhadran ``Gradient Steepness Metrics Using Extended Baum-Welch Transformations for Universal Pattern Recognition Tasks\" , submitted for ICASSP 2008.\r\n\r\nR. Schluter, W. Macherey, B. Muler and H. Ney, \"Comparison of discriminative training criteria and optimization methods for speech recognition\", Speech Communication, Vol. 34, pp.287-310, 2001.\r\n\r\nV. Valtchev , P.C. Woodland and S. J. Young, \"Lattice-based Discriminative Training for Large Vocabulary Speech Recognition Systems\", Speech Communication, Vol. 22, pp. 303-314, 1996.\r\n\r\n" relation_type: [] relation_uri: [] reportno: ~ rev_number: 29 series: ~ source: ~ status_changed: 2008-01-15 23:56:42 subjects: - comp-sci-speech succeeds: ~ suggestions: ~ sword_depositor: ~ sword_slug: ~ thesistype: ~ title: Adapted Extended Baum-Welch transformations type: preprint userid: 6589 volume: ~