Implicit Trajectory Modeling Using Temporally Varying Model Parameters

--------------------------------------------------------------------
              ***Joint Seminar***
--------------------------------------------------------------------
      Department of Computer Science and Engineering, HKUST
            Human Language Technology Center, HKUST
          IEEE Hong Kong Chapter of Signal Processing
--------------------------------------------------------------------

Speaker:        Dr. Khe Chai SIM
                National University of Singapore

Title:          "Implicit Trajectory Modeling Using Temporally
                Varying Model Parameters"

Date:           Friday, 16 March 2012

Time:           2:00pm - 3:00pm

Venue:          Room 5510 (via lifts 25/26), HKUST


Abstract:

Hidden Markov Model (HMM) is widely used to represent acoustic units in
automatic speech recognition systems. One of the major limitations of HMM
is the conditional independence assumption of the observation vector given
its state. This leads to a poor trajectory model. Although many explicit
trajectory modeling techniques have been proposed and studied in the past,
the use of dynamic parameters remains the most popular way to circumvent
the problem. Semi-parametric trajectory models have been proposed to
implicitly model the trajectory using temporally varying model parameters.
These parameters are modeled using a linear regression of some basis
structure with temporally varying regression weights. Initial work has
focused on modeling the temporally varying mean (fMPE) and precision
matrices (pMPE). Recently, Temporally Varying Weight Regression (TVWR) has
been proposed to model the component weights of the Gaussian Mixture Model
(GMM). As a result, TVWR is able to model non-stationary distribution
within the HMM states, yielding an implicit trajectory model. Experimental
results show that TVWR outperforms standard HMM systems using both maximum
likelihood and discriminative training schemes. Furthermore, it is also
possible to combine TVWR and fMPE to yield further improvements.


*******************
Biography:

Dr. Khe Chai Sim is an Assistant Professor at the School of Computing,
National University of Singapore. He received the B.A. and M.Eng degrees
in Electrical and Information Sciences from the University of Cambridge,
England in 2001. He worked on the API for Hidden Markov Model Toolkit
(HTK) (a.k.a. the ATK) under the supervision of Prof. Steve Young. He was
awarded the Gates Cambridge Scholarship and completed his M.Phil
dissertation in 2002 under the supervision of Dr. Mark Gales. He joined
the Machine Intelligence Laboratory, Cambridge University Engineering
Department in the same year as a research student, supervised by Dr. Mark
Gales. He received his Ph.D degree in 2006. He is also an alumni of
Churchill College. His main research interest is in statistical pattern
classification and acoustic modelling for automatic speech recognition. He
worked on the DARPA funded projects, EARS and GALE, between 2002-2006. He
was also in the IIR team which participated in the NIST 2007 Language
recognition Evaluation and the NIST 2008 Speaker Recognition Evaluation.