|
Jorge
Silva
Ph.
D. Candidate:Adviser
Shrikanth
S. Narayanan
Viterbi
Doctoral Fellowship 2007-2008
Simon
Ramo Scholarship 2007-2008
Speech
Analysis and Interpretation Laboratory (SAIL)
Signal and Image Processing Institute (SIPI)
Department of Electrical Engineering
University of Southern California, Viterbi School of Engineering
(USC)
3710
S. McClintock Ave, RTH 320
Los Angeles, CA 90089, U.S.A
Tel: (213) 740 3477
Currently
holding a full-time Faculty position at
Faculty of Physical and Mathematical
Sciences (fcfm)
Department of Electrical Engineering
(DIE)
UNIVERSITY
OF CHILE
Research
Statement:
My
general research interest is to understand the role of Information
Theory ---information quantities, problem formulations and
theoretical results --- in Signal Processing problems with
particular focus on pattern recognition. My recent work can
be divided in two directions: first in defining statistical
discrimination measures for hidden Markov models (HMMs)
with applications to speech recognition problems and secondly
in formulating the notion of optimal signal representation
for classification by studying conceptual connection with
rate distortion theory for lossy compression and structural
risk minimization (SRM) in statistical learning theory.
In this last direction, I have been working in particularizing
optimal signal representation problems
for tree-structured feature representations ---Wavelet Packets
(WPs )--- by finding
conceptual connection with well studied and understood problems:
regression and classification trees (CART, Breiman
et al) and tree-structured
vector quantization.
My
current research
interests
include: signal representation for pattern recognition; speech
recognition; vector quantization for lossy compression and
classification; tree-structured representations for classification
and signal representation (classification trees (CART)
and Wavelet packets (WPs)); distributive learning,
sequential detection and sensor networks.
My
general thematic interests include: probability theory and
stochastic processes, information theory, statistical signal
processing, wavelet and multiresolution analysis and statistical
learning theory.
Short
Bibliography:
Jorge
Silva was born in Santiago, Chile, in 1977. He received
the B.Sc. and engineering degree in electrical engineering
with highest distinction from the University of Chile, Santiago,
in 2002 and Master of Science in electrical engineering from
the University of Southern California, in Spring 2005.
From
2000 to 2003, he was a research assistant in the Speech
Processing and Transmission Laboratory(LPTV),
Department
of Electrical Engineering at University of Chile,
where he worked on robust speech recognition, state duration
distribution analysis and developing speaker adaptation techniques.
During 2002-2003, he was a part-time lecturer at the Department
of Electrical Engineering from University of Chile, where
he taught undergraduate courses in digital signal processing
and computer architecture. During summer 2005 he was research
intern at the Speech Research Group, Microsoft Corporation,
collaborating in various aspect of spoken document retrieval.
Since
Fall 2003 he has been Ph. D. student in the Department
of Electrical Engineering at the University of
Southern California (USC) and a member of the Speech
Analysis and Interpretation Laboratory (SAIL)
at USC.
Recent
Papers:
J.
Silva and S. Narayanan, "Upper
Bound Kullback-Leibler Divergence for Transient Hidden Markov
Models,"
in Press, IEEE Transactions on Signal Processing.[Abstract,
pdf]
J.
Silva and S. Narayanan, "Minimum Probability of Error
Signal Representation," in IEEE
International Workshop on Machine Learning for Signal Processing,
Thessaloniki, Greece, 2007.[pdf]
Ciprian
Chelba, Jorge Silva and Alex Acero, “Soft Indexing
of Speech Content for Search in Spoken Documents,” Elsevier
Computer Speech and Language, vol. 21, issue 3, pp. 458-478,
July 2007. [Abstract
, pdf]
J.
Silva and S. Narayanan, "Universal Consistency of
Data-Driven Partitions for Divergence Estimation,"
in IEEE
International Symposium on Information Theory, Nice, France,
2007.
[pdf]
Full
List....
More
Details:
Resumen
Formation
|
Universidad
de Chile
|
University
of Southern California
|
last
update, February 2008
|
