An Abridged Statement of Research Interests
Theory and Applications of Machine Learning
Muhammad Adil Raja∗
December 22, 2014
My research interests fall in the general area of theory and applications of machine learn-
ing. Machine learning is a sub-field of artificial intelligence that is concerned with design and
development of algorithms that allow computers to automatically create solutions to user
specified problems based on empirical data [Mitchell, 1997]. More specifically, my research
falls under two themes: (a) to develop a thorough understanding of the theoretical concepts
of various methods, algorithms and sub-domains of machine learning; and (b) to effectively
apply these ideas and concepts for solving various real-world scientific problems. Machine
learning offers solutions to many complex, computationally hard real-world problems that
are other otherwise difficult to solve. As more intricate scientific and engineering problems
have emerged the need to develop sophisticated machine learning algorithms as well as to
refine the existing algorithms is also required. To this end, I have an insatiable desire for
addressing the first objective stated earlier. I wish to learn more about various domains and
algorithms of machine learning that primarily include genetic algorithms and programming,
support vector machines, clustering techniques, Bayesian networks, Markov modeling, rein-
forcement learning, unsupervised learning and also the theoretical concepts of computational
learning theory. I also have a keen interest in enhancing my knowledge about mathemat-
ics. In parallel to this I wish to apply my knowledge of machine learning to (a) real-world
problem(s).
I started my PhD studies in February 2005 in University of Limerick in Ireland under
the supervision of Dr. Colin Flanagan and I was awarded the PhD degree in August 2008.
It took me precisely three years and three months to accomplish my PhD studies. After
that I also had a chance to work as a postdoctoral researcher with Orange Labs, France
Telecom R&D, Lannion, France. The topic I have been actively studying and addressing
so far is the speech quality estimation problem. Speech quality, as perceived by the users of
Voice over Internet Protocol (VoIP) telephony, is critically important to the uptake of this
service. VoIP quality can be degraded by network layer problems (delay, jitter, packet loss).
The research presented methods for real-time, non-intrusive speech quality estimation for
VoIP that emulated the subjective listening quality measures based on Mean Opinion Scores
(MOS). MOS provides the numerical indication of perceived quality of speech. A Genetic
∗
An Abridged Statement of Research Interests by Muhammad Adil Raja is licensed under a Creative
Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. cbnd
1
Programming (GP) [Koza, 1992] based symbolic regression approach was adopted to derive
speech quality estimation models. The results compared favorably with various standards
proposed by International Telecommunications Union-Telecommunication Standardization
(ITU-T). Moreover, the models are suitable for real-time speech quality estimation of VoIP.
The research resulted in numerous outstanding publications [Raja et al., 2006, Raja et al.,
2007,Raja and Flanagan, 2008,Raja et al., 2008]. Among these, [Raja et al., 2007] was also
nominated for best paper award by the EuroGP 2007 conference committee 1
. Honorable
mentions were awarded twice for the 2007 and 2008 Hummies awards for human-competitive
results produced by genetic and evolutionary computation 2
. Moreover, during my PhD
I also had a chance to collaborate with a center of excellence namely Biocomputing and
Developmental Systems (BDS) Group, based in University of Limerick, Ireland and headed
by Dr. Conor Ryan. The group pioneers in research concerning evolutionary algorithms
and their applications to real world problems. Recently, During my postdoctoral research
in France Telecom I also had a chance to build industry academia liaison between France
Telecom and the BDS group. At France Telecom I also found an excellent match of my
research work in the industry.
While working on speech quality estimation I also performed a thorough literature review
of various other related fields which include speech processing, speech coding, speech synthe-
sis, automatic speech recognition, auditory scene analysis and the various machine learning
methods that may be applied to solve practical problems that arise in these domains. I also
studied other elementary subjects such as digital signal and image processing, statistics, lin-
ear algebra and differential calculus during my PhD studies. As part of my research I was also
heavily involved in software development at different phases of my studies. I learned various
programming languages and acquired new techniques for handling large amounts of data.
I also wrote a few rather huge applications for the sake of fun and to satiate my curiosity
about my understanding of the working and implementation of algorithms. This includes:
development of a genetic algorithms software for numerical optimization, development of a
genetic programming system for symbolic regression, development of a grammatical evolu-
tion system for optimization. I also gained hands on experience about working with, and
development of, various speech processing tools used in problems such as speech coding and
recognition.
Apart from this, over the past years, I have been avidly reading about various other
researchable questions in the general area of machine learning or its applications. These
include diverse problem domains such as communication networks, digital hardware design,
computational neuroscience, digital signal processing, artificial reality, artificial conscious-
ness, psychology and computer aided diagnosis. At some stage I would like to actively
pursue some problems in these domains.
Overall my personal experience about being a researcher, or of doing research, has been
a very enjoyable one. Doing research has ever since been more of a hobby for me than a
professional choice. For me it is the best way to address my euphoria and curiosity about
the world’s body of knowledge.
1
http://www.informatik.uni-trier.de/ ley/db/conf/eurogp/eurogp2007.html
2
http://www.genetic-programming.org/hc2007/cfe2007.html
2
References
[Koza, 1992] Koza, J. R. (1992). Genetic Programming: On the Programming of Computers
by Means of Natural Selection. MIT Press, Cambridge, MA, USA.
[Mitchell, 1997] Mitchell, T. (1997). Machine Learning. McGraw Hill, New York.
[Raja et al., 2006] Raja, A., Azad, R. M. A., Flanagan, C., Picovici, D., and Ryan, C.
(2006). Non-intrusive quality evaluation of voip using genetic programming. In First
International Conference on Bio Inspired Models of Network, Information and Computer
Systems, volume 4, pages 2573–2577.
[Raja et al., 2007] Raja, A., Azad, R. M. A., Flanagan, C., and Ryan, C. (2007). Real-
time, non-intrusive evaluation of VoIP. In Ebner, M., O’Neill, M., Ek´art, A., Vanneschi,
L., and Esparcia-Alc´azar, A. I., editors, Proceedings of the 10th European Conference on
Genetic Programming, volume 4445 of Lecture Notes in Computer Science, pages 217–228,
Valencia, Spain. Springer.
[Raja et al., 2008] Raja, A., Azad, R. M. A., Flanagan, C., and Ryan, C. (2008). A method-
ology for deriving VoIP equipment impairment factors for a mixed NB/WB context. IEEE
Transactions on Multimedia, 10(6):1046–1058.
[Raja and Flanagan, 2008] Raja, A. and Flanagan, C. (2008). Real-time, non-intrusive
speech quality estimation: A signal-based model. In O’Neill, M., Vanneschi, L., Gustafson,
S., Esparcia Alcazar, A. I., De Falco, I., Della Cioppa, A., and Tarantino, E., editors, Pro-
ceedings of the 11th European Conference on Genetic Programming, EuroGP 2008, volume
4971 of Lecture Notes in Computer Science, pages 37–48, Naples. Springer.
3

An Abridged Version of My Statement of Research Interests

  • 1.
    An Abridged Statementof Research Interests Theory and Applications of Machine Learning Muhammad Adil Raja∗ December 22, 2014 My research interests fall in the general area of theory and applications of machine learn- ing. Machine learning is a sub-field of artificial intelligence that is concerned with design and development of algorithms that allow computers to automatically create solutions to user specified problems based on empirical data [Mitchell, 1997]. More specifically, my research falls under two themes: (a) to develop a thorough understanding of the theoretical concepts of various methods, algorithms and sub-domains of machine learning; and (b) to effectively apply these ideas and concepts for solving various real-world scientific problems. Machine learning offers solutions to many complex, computationally hard real-world problems that are other otherwise difficult to solve. As more intricate scientific and engineering problems have emerged the need to develop sophisticated machine learning algorithms as well as to refine the existing algorithms is also required. To this end, I have an insatiable desire for addressing the first objective stated earlier. I wish to learn more about various domains and algorithms of machine learning that primarily include genetic algorithms and programming, support vector machines, clustering techniques, Bayesian networks, Markov modeling, rein- forcement learning, unsupervised learning and also the theoretical concepts of computational learning theory. I also have a keen interest in enhancing my knowledge about mathemat- ics. In parallel to this I wish to apply my knowledge of machine learning to (a) real-world problem(s). I started my PhD studies in February 2005 in University of Limerick in Ireland under the supervision of Dr. Colin Flanagan and I was awarded the PhD degree in August 2008. It took me precisely three years and three months to accomplish my PhD studies. After that I also had a chance to work as a postdoctoral researcher with Orange Labs, France Telecom R&D, Lannion, France. The topic I have been actively studying and addressing so far is the speech quality estimation problem. Speech quality, as perceived by the users of Voice over Internet Protocol (VoIP) telephony, is critically important to the uptake of this service. VoIP quality can be degraded by network layer problems (delay, jitter, packet loss). The research presented methods for real-time, non-intrusive speech quality estimation for VoIP that emulated the subjective listening quality measures based on Mean Opinion Scores (MOS). MOS provides the numerical indication of perceived quality of speech. A Genetic ∗ An Abridged Statement of Research Interests by Muhammad Adil Raja is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. cbnd 1
  • 2.
    Programming (GP) [Koza,1992] based symbolic regression approach was adopted to derive speech quality estimation models. The results compared favorably with various standards proposed by International Telecommunications Union-Telecommunication Standardization (ITU-T). Moreover, the models are suitable for real-time speech quality estimation of VoIP. The research resulted in numerous outstanding publications [Raja et al., 2006, Raja et al., 2007,Raja and Flanagan, 2008,Raja et al., 2008]. Among these, [Raja et al., 2007] was also nominated for best paper award by the EuroGP 2007 conference committee 1 . Honorable mentions were awarded twice for the 2007 and 2008 Hummies awards for human-competitive results produced by genetic and evolutionary computation 2 . Moreover, during my PhD I also had a chance to collaborate with a center of excellence namely Biocomputing and Developmental Systems (BDS) Group, based in University of Limerick, Ireland and headed by Dr. Conor Ryan. The group pioneers in research concerning evolutionary algorithms and their applications to real world problems. Recently, During my postdoctoral research in France Telecom I also had a chance to build industry academia liaison between France Telecom and the BDS group. At France Telecom I also found an excellent match of my research work in the industry. While working on speech quality estimation I also performed a thorough literature review of various other related fields which include speech processing, speech coding, speech synthe- sis, automatic speech recognition, auditory scene analysis and the various machine learning methods that may be applied to solve practical problems that arise in these domains. I also studied other elementary subjects such as digital signal and image processing, statistics, lin- ear algebra and differential calculus during my PhD studies. As part of my research I was also heavily involved in software development at different phases of my studies. I learned various programming languages and acquired new techniques for handling large amounts of data. I also wrote a few rather huge applications for the sake of fun and to satiate my curiosity about my understanding of the working and implementation of algorithms. This includes: development of a genetic algorithms software for numerical optimization, development of a genetic programming system for symbolic regression, development of a grammatical evolu- tion system for optimization. I also gained hands on experience about working with, and development of, various speech processing tools used in problems such as speech coding and recognition. Apart from this, over the past years, I have been avidly reading about various other researchable questions in the general area of machine learning or its applications. These include diverse problem domains such as communication networks, digital hardware design, computational neuroscience, digital signal processing, artificial reality, artificial conscious- ness, psychology and computer aided diagnosis. At some stage I would like to actively pursue some problems in these domains. Overall my personal experience about being a researcher, or of doing research, has been a very enjoyable one. Doing research has ever since been more of a hobby for me than a professional choice. For me it is the best way to address my euphoria and curiosity about the world’s body of knowledge. 1 http://www.informatik.uni-trier.de/ ley/db/conf/eurogp/eurogp2007.html 2 http://www.genetic-programming.org/hc2007/cfe2007.html 2
  • 3.
    References [Koza, 1992] Koza,J. R. (1992). Genetic Programming: On the Programming of Computers by Means of Natural Selection. MIT Press, Cambridge, MA, USA. [Mitchell, 1997] Mitchell, T. (1997). Machine Learning. McGraw Hill, New York. [Raja et al., 2006] Raja, A., Azad, R. M. A., Flanagan, C., Picovici, D., and Ryan, C. (2006). Non-intrusive quality evaluation of voip using genetic programming. In First International Conference on Bio Inspired Models of Network, Information and Computer Systems, volume 4, pages 2573–2577. [Raja et al., 2007] Raja, A., Azad, R. M. A., Flanagan, C., and Ryan, C. (2007). Real- time, non-intrusive evaluation of VoIP. In Ebner, M., O’Neill, M., Ek´art, A., Vanneschi, L., and Esparcia-Alc´azar, A. I., editors, Proceedings of the 10th European Conference on Genetic Programming, volume 4445 of Lecture Notes in Computer Science, pages 217–228, Valencia, Spain. Springer. [Raja et al., 2008] Raja, A., Azad, R. M. A., Flanagan, C., and Ryan, C. (2008). A method- ology for deriving VoIP equipment impairment factors for a mixed NB/WB context. IEEE Transactions on Multimedia, 10(6):1046–1058. [Raja and Flanagan, 2008] Raja, A. and Flanagan, C. (2008). Real-time, non-intrusive speech quality estimation: A signal-based model. In O’Neill, M., Vanneschi, L., Gustafson, S., Esparcia Alcazar, A. I., De Falco, I., Della Cioppa, A., and Tarantino, E., editors, Pro- ceedings of the 11th European Conference on Genetic Programming, EuroGP 2008, volume 4971 of Lecture Notes in Computer Science, pages 37–48, Naples. Springer. 3