INTERESTS
Unsupervised learning techniques for natural language
processing. Induction of syntactic and semantic structure of language
from text corpora and from interaction with language users.
Computerised natural-language translation.
Applying learning algorithms to speech recognition, speech
synthesis, and speech understanding systems.
Learning and information processing algorithms motivated from
neuroanatomy and physiological psychology.
Parallel Computer Architecture and Parallel Algorithm design.
Distributed applications and distributed linguistic data acquistion
using the internet.
EDUCATION
B.Sc. (Hons) - Bachelor of Science in Psychology
with First Class Honours, University of Otago, Dunedin, New Zealand (1984).
Substantial Minor in Computer Science. G.P.A 4.0
Computer Science PhD program at
Carnegie Mellon 1986 to
present. Major Area:
Connectionist A.I. Minor area: Parallel Computer Architecture.
MS - Master of Science in Computer Science (Carnegie Mellon, May 1990).
PhD - Expected March 1995. Thesis Topic: Learning Models of Speaker Variation.
Languages Spoken: English, French. Read: Spanish.
EMPLOYMENT EXPERIENCE
7/91-9/91
IBM T.J.WATSON RESEARCH CENTER, YORKTOWN NY.
Evaluated the
suitability of the GF11 SIMD architecture for molecular dynamics
simulations. Designed and implemented a fast general message passing communication system for GF11 using its Benes network. Manager: Dr. Manoj Kumar.
5/91-7/91
CENTRE NATIONAL D'ÉTUDES DES TÉLÉCOMMUNICATIONS, LANNION, FRANCE.
Research on using connectionist learning techniques to characterise speaker voice variability. Manager: Dr. Christel Sorin.
5/90-8/90
IBM T.J.WATSON RESEARCH CENTER, YORKTOWN NY.
Designed and implemented
parallel LU decomposition and matrix inversion routines for GF11. Extended programing tools to handle communications requirements for this task. Used RS6000s to work on neural network recognition of structure in sequential inputs.
Manager: Dr. Manoj Kumar.
5/89-8/89
IBM T.J.WATSON RESEARCH CENTER, YORKTOWN NY.
Designed and
implemented a backpropagation neural network simulator for GF11. The
resulting program was the fastest existing backpropagation simulator.
Manager: Dr. David A George.
11/84-8/86
TEXT UNDERSTANDING SYSTEMS, DUNEDIN, NEW ZEALAND
Senior Programmer/Analyst on a project to produce an automatic
natural-language translation System for micro-computers, based on
SYSTRAN. Was responsible for directing a team of 8 programmers, and for working with the company's linguists on
system design. Director: Dr. Peter Toma
11/83-3/84
UNIVERSITY OF OTAGO, DUNEDIN, NEW ZEALAND
Research Assistant. Worked on development of an automated system for
staging sleep EEGs. Also responsible for recording and manual analysis
of EEGs. Supervisor Dr. David. F. Marks.
11/82-3/83
UNIVERSITY OF OTAGO, DUNEDIN, NEW ZEALAND
Research Assistant. Designed and implemented a raster
graphics library for Otago University's computer graphics project. Also worked
on hardware for graphical input devices and computer interconnect. Supervisor: Dr. Geoffrey Wyvill
11/78-3/79
THE WELLCOME MEDICAL RESEARCH INSTITUTE, DUNEDIN, NEW ZEALAND
Programmer. Implementation of automated system for analysing hand grip
strengths for arthritic patients. Supervisor Dr. Myers.
TEACHING EXPERIENCE
2/84-11/84
OTAGO UNIVERSITY, PSYCHOLOGY.
Laboratory tutor for
Introduction to Experimental methods. Taught and graded one
half of a full year laboratory course.
9/88-12/84
CARNEGIE MELLON UNIVERSITY, COMPUTER SCIENCE.
Teaching
Assistant for Introductory AI course. Prepared and presented two
lectures, designed and graded homework assignments and exam
questions.
PUBLICATIONS
10/94
Fischer, Michael., Meyer, Michael., and Witbrock, Michael. ``User
Extensibility in Amiga Mosaic'', Proceedings
of the Second International World Wide Web (WWW) Conference '94: Mosaic and the
Web, October 1994. In Press.
3/92
Witbrock, Michael and Haffner, Patrick. ``Rapid Connectionist Speaker Adaptation'', Proceedings of the IEEE 1992 International Conference on Acoustics,Speech, and Signal Processing, March, 1992.
12/91
Witbrock, Michael and Zagha, Marco. ``Back-Propagation Learning on
the IBM GF11'', Chapter in Przytula, K.W., and Prasanna Kumar, V.K.
Parallel Digital Implementations of Neural Networks PTR Prentice
Hall. Englewood Cliffs. 1993.
12/89
Witbrock, Michael and Zagha, Marco. ``An Implementation of Back-Propagation Learning on GF11, a Large SIMD Parallel Computer'', Parallel Computing 14 (1990) pp 329-346, North-Holland, Amsterdam
6/89
Franzini, M.A., Witbrock. M.J. and Lee, K.F. ``Speaker-Independent Recognition of Connected Utterances Using
Recurrent and Non-recurrent Neural Networks'', Proceedings of the I.E.E.E. International Joint Conference on Neural Networks., June, 1989.
5/89
Franzini, M.A., Witbrock, M.J. and Lee, K. ``Connectionist Approaches to Continuous
Speech Recognition'' Proceedings of the IEEE 1989 International Conference on Acoustics,
Speech, and Signal Processing, May, 1989.
9/88
Lang, K.J. and Witbrock, M.J. ``Learning to Tell Two Spirals Apart'' in
Touretzky, D.S., Hinton, G.E., and Sejnowski, T. (Eds) Proceedings of
1988 Connectionist Models Summer School