| | -----------------------------------------------------
- INTERSPEECH 2014 - SINGAPORE -
- September 14-18, 2014 -
- http://www.interspeech2014.org -
-----------------------------------------------------
ISCA, COLIPS and the organizing Committee of INTERSPEECH 2014
are proud to announce that INTERSPEECH 2014 will feature
five plenary talks by internationally renowned experts.
- keynote speech
by the ISCA Medallist 2014
- 'Decision Learning in Data Science:
Where John Nash Meets Social Media'
by Professor K. J. Ray Liu
- 'Language Diversity: Speech Processing In A Multi-Lingual Context'
by Dr. Lori Lamel
- 'Sound Patterns In Language'
by Professor William Shi-Yuan WANG 王士元
- 'Achievements and Challenges of Deep Learning
From Speech Analysis And Recognition To Language
And Multimodal Processing'
by Dr. Li DENG
Details of the keynote speeches and biographies of the presenters are given below.
Looking forward to welcome you in Singapore,
the organizing committee
************************************************************************
* On Monday, 15th of September *
************************************************************************
The ISCA Medallist 2014 will give a keynote speech.
The name of the Medallist and subject of the talk will be
disclosed on the first day of INTERSPEECH 2014.
************************************************************************
* On Tuesday morning, 16th of September *
************************************************************************
Professor K. J. Ray Liu
Department of Electrical and Computer Engineering
University of Maryland, College Park
will give a presentation on:
'Decision Learning in Data Science: Where John Nash Meets Social Media'
Abstract
With the increasing ubiquity and power of mobile devices,
as well as the prevalence of social media, more and more
activities in our daily life are being recorded, tracked,
and shared, creating the notion of “social media”.
Such abundant and still growing real life data, known as
“big data”, provide a tremendous research opportunity in many fields.
To analyze, learn and understand such user-generated big data,
machine learning has been an important tool and various
machine learning algorithms have been developed.
However, since the user-generated big data is
the outcome of users’ decisions, actions and their socio-economic
interactions, which are highly dynamic, without considering users’
local behaviours and interests, existing learning approaches
tend to focus on optimizing a global objective function at
the macroeconomic level, while totally ignore users’ local
decisions at the micro-economic level. As such there is a growing
need in bridging machine/social learning with strategic decision
making, which are two traditionally distinct research disciplines,
to be able to jointly consider both global phenomenon and local
effects to understand/model/analyze better the newly arising
issues in the emerging social media. In this talk, we present
the notion of “decision learning” that can involve users's
behaviours and interactions by combining learning with strategic
decision making.
We will discuss some examples from social media with real data to
show how decision learning can be used to better analyze users’
optimal decision from a user’ perspective as well as design a
mechanism from the system designer’s perspective
to achieve a desirable outcome.
Biography of the speaker
Dr. K. J. Ray Liu was named a Distinguished Scholar-Teacher
of University of Maryland in 2007, where he is Christine Kim
Eminent Professor of Information Technology.
He leads the Maryland Signals and Information Group conducting
research encompassing broad areas of signal processing and
communications with recent focus on cooperative communications,
cognitive networking, social learning and decision making,
and information forensics and security. Dr. Liu has received
numerous honours and awards including IEEE Signal Processing
Society 2009 Technical Achievement Award and various best paper
awards from IEEE Signal Processing, Communications, and Vehicular
Technology Societies, and EURASIP. A Fellow of the IEEE and AAAS,
he is recognized by Thomson Reuters as an ISI Highly Cited
Researcher.
Dr. Liu was the President of IEEE Signal Processing Society,
the Editor-in-Chief of IEEE Signal Processing Magazine and
the founding Editor-in-Chief of EURASIP Journal on Advances
in Signal Processing. Dr. Liu also received various research
and teaching recognitions from the University of Maryland,
including Poole and Kent Senior Faculty Teaching Award,
Outstanding Faculty Research Award, and Outstanding Faculty
Service Award, all from A. James Clark School of Engineering;
and Invention of the Year Award (three times)
from Office of Technology Commercialization.
************************************************************************
* On Tuesday afternoon, 16th of September *
************************************************************************
Dr. Lori Lamel
Senior Research scientist (DR1), LIMSI-CNRS
will give a presentation on
'Language Diversity: Speech Processing In A Multi-Lingual Context'
Abstract
Speech processing encompasses a variety of technologies
that automatically process speech for some downstream processing.
These technologies include identifying the language or dialect
spoken, the person speaking, what is said and how it is said.
The downstream processing may be limited to a transcription or
to a transcription enhanced with additional meta-data, or may
be used to carry out an action or interpreted within a spoken
dialogue system or more generally for analytics. With the
availability of large spoken multimedia or multimodal data there is
growing interest in using such technologies to provide structure
and random access to particular segments. Automatic tools can also
serve to annotate large corpora for exploitation in linguistic
studies of spoken language, such as acoustic-phonetics,
pronunciation variation and diachronic evolution,
permitting the validation of hypotheses and models.
In this talk I will present some of my experience with speech
processing in multiple languages, drawing upon progress in the
context of several research projects, most recently the Quaero
program and the IARPA Babel program, both of which address the
development of technologies in a variety of languages, with the aim
to some highlight recent research directions and challenges.
Biography of the speaker
I am a senior research scientist (DR1) at the CNRS, which I joined as
a permanent researcher at LIMSI in October 1991.
I received my Ph.D. degree in Electrical Engineering and Computer Science
in May 1988 from the Massachusetts Institute of Technology.
My research activities focus on large vocabulary speaker-
independent, continuous speech recognition in multiple languages
with a recent focus on low-resourced languages; lightly and
unsupervised acoustic model training methods; studies in acoustic-
phonetics; lexical and pronunciation modelling. I contributed to
the design, and realization of large speech corpora (TIMIT, BREF,
TED). I have been actively involved in the research projects, most
recently leading the activities on speech processing in the OSEO
Quaero program, and I am currently co-principal investigator for
LIMSI as part of the IARPA Babel Babelon team led by BBN.
I served on the Steering committee for Interspeech 2013 as
co-technical program chair along with Pascal Perrier, and I am now
serving on the Technical Program Committee of Interspeech 2014.
************************************************************************
* On Wednesday, 17th of September *
************************************************************************
Professor William Shi-Yuan WANG 王士元
Centre for Language and Human Complexity,
Chinese University of Hong Kong
Professor Emeritus, University of California at Berkeley
Honorary Professor, Peking University
Academician, Academia Sinica
will give a presentation about
'Sound Patterns In Language'
Abstract
In contrast to other species, humans are unique in having developed
thousands of diverse languages which are not mutually
intelligible. However, any infant can learn any language with ease,
because all languages are based upon common biological
infrastructures of sensori-motor, memorial, and cognitive
faculties. While languages may differ significantly in the sounds
they use, the overall organization is largely the same.
It is divided into a discrete segmental system for building words
and a continuous prosodic system for expressing, phrasing,
attitudes, and emotions. Within this organization, I will discuss a
class of languages called 'tone languages', which makes special use
of F0 to build words. Although the best known of these is Chinese,
tone languages are found in many parts of the world, and operate on
different principles. I will also comment on relations between
sound patterns in language and sound patterns in music, the two
worlds of sound universal to our species.
Biography of the speaker
William S-Y. Wang received his early schooling in China, and his
PhD from the University of Michigan. He was appointed
Professor of Linguistics at the University of California at
Berkeley in 1965, and taught there for 30 years.
Currently he is in the Department of Electronic Engineering and in
the Department of Linguistics and Modern Languages of the Chinese
University of Hong Kong, and Director of the newly established
Joint Research Centre for Language and Human Complexity. His
primary interest is the evolution of language from a multi-
disciplinary perspective.
************************************************************************
* On Thursday, 18th of September *
************************************************************************
Dr. Li DENG
Principal Researcher and Research Manager
Deep Learning Technology Centre,
Microsoft Research, Redmond, USA
will give a presentation on the
'Achievements and Challenges of Deep Learning
From Speech Analysis And Recognition To Language And Multimodal Processing'
Abstract
Artificial neural networks have been around for over half a century
and their applications to speech processing have been almost as
long, yet it was not until year 2010 that their real impact had
been made by a deep form of such networks, built upon part of the
earlier work on (shallow) neural nets and (deep) graphical models
developed by both speech and machine learning communities. This
keynote will first reflect on the path to this transformative
success, sparked by speech analysis using deep learning methods
on spectrogram-like raw features and then progressing rapidly to
speech recognition with increasingly larger vocabularies and scale.
The role of well-timed academic-industrial collaboration will be
highlighted, so will be the advances of big data, big compute, and
the seamless integration between the application-domain knowledge
of speech and general principles of deep learning. Then, an
overview will be given on sweeping achievements of deep learning in
speech recognition since its initial success in 2010 (as well as in
image recognition and computer vision since 2012). Such
achievements have resulted in across-the-board, industry-wide
deployment of deep learning. The final part of the talk will look
ahead towards stimulating new challenges of deep learning ---
making intelligent machines capable of not only hearing (speech)
and seeing (vision), but also of thinking with a “mind”; i.e.
reasoning and inference over complex, hierarchical relationships
and knowledge sources that comprise a vast number of entities
and semantic concepts in the real world based in part on multi-
sensory data from the user. To this end, language and multimodal
processing --- joint exploitation and learning from text,
speech/audio, and image/video --- is evolving into a new frontier
of deep learning, beginning to be embraced by a mixture of research
communities including speech and spoken language processing,
natural language processing, computer vision, machine learning,
information retrieval, cognitive science, artificial intelligence,
and data/knowledge management. A review of recent published studies
will be provided on deep learning applied to selected language and
multimodal processing tasks, with a trace back to the relevant
early connectionist modelling and neural network literature and
with future directions in this new exciting deep learning frontier
discussed and analyzed.
Biography of the speaker
Li Deng received Ph.D. from the University of Wisconsin-Madison.
He was a tenured professor (1989-1999) at the University of
Waterloo, Ontario, Canada, and then joined Microsoft Research,
Redmond, where he is currently a Principal Research Manager of its
Deep Learning Technology Centre.
Since 2000, he has also been an affiliate full professor at the
University of Washington, Seattle, teaching computer speech
processing. He has been granted over 60 US or international
patents, and has received numerous awards and honours
bestowed by IEEE, ISCA, ASA, and Microsoft including the latest
IEEE SPS Best Paper Award (2013) on deep neural nets for speech
recognition. He authored or co-authored 4 books including the
latest one on Deep Learning: Methods and Applications. He is a
Fellow of the Acoustical Society of America, a Fellow of the IEEE,
and a Fellow of the ISCA. He served as the Editor-in-Chief
for IEEE Signal Processing Magazine (2009-2011), and currently as
Editor-in-Chief for IEEE Transactions on Audio, Speech and Language
Processing. His recent research interests and activities have been
focused on deep learning and machine intelligence applied to
large-scale text analysis and to speech/language/image
multimodal processing, advancing his earlier work with
collaborators on speech analysis and recognition using deep neural
networks since 2009.
|
