Dr. M.K. Ahilan


Senior Lecturer Grade I
Department of Electrical and Electronic Engineering


B. Sc. Eng (Peradeniya), PhD(QUT,Australia)
Senior Member of the IEEE, AMIE(SL), MCS(SL)


Research interests : Applying machine learning to solve business problems and help organizations make data-driven decisions.


  ahilan.eng@gmail.com     +94 21 206 0160


Dr. M.K. Ahilan joined the Department of Electrical and Electronic Engineering, Faculty of Engineering, University of Jaffna, Sri Lanka in 2017 as a senior lecturer. He is also working as visiting researcher at Speech, Audio, Image and Video Technologies (SAIVT) lab, Queensland University of Technology, Australia. He is holding Australian and Sri Lankan Citizenships since 2016.

He received B.Sc Eng Degree with first class honours (3.80/4.00) from University of Peradeniya, Sri Lanka in 2008 and PhD from Queensland University of Technology, Australia in 2014.

Ahilan worked as machine learning and speech processing researcher at SAIVT lab, Queensland University of Technology, Australia from Oct 2014 to July 2016. He also worked as visiting researcher at ATVS lab, Autonomous University of Madrid, Spain from Nov 2012 to Apr 2013. From September 2008 till July 2010, he worked as Business Development and Technical Support Engineer at Sri Lanka Telecom (PLC) Head Quarters, Colombo.


(1) Senior Member of the IEEE
(2) Associate Member of the IESL
(3) Professional Member of CSSL


  • ICT Researcher of the Year by SEARCC Global ICT Awards (2018).
  • ICT researcher of the year award by Computer Society of Sri Lanka (CSSL) (2018).
  • Australasian Speech Science and Technology Association (ASSTA) new researcher award, 2014.
  • Australasian Speech Science and Technology Association (ASSTA) travel award, 2013.


  • NVIDIA GPU grant (2018).
  • University research grant (2018 - 2019)
  • India Science & Research Fellowship award (2018 - 2019).
  • European Commission Marie Curie ITN BBfor2 network sponsored internship program, 2012.
  • Cooperative Research Centre for Advanced Automotive Technology AutoCRC postgraduate scholarship, 2010.
  • AHEAD - Research, Innovation and Commercialization Grants - 36 million ( 2020 - 2022)

  • Applying wide range of machine learning and deep learning algorithms to different speech signal processing and computer vision applications such as speech recognition, speaker diarization, speaker recognition, and face recognition.
  • Applying machine learning to solve business problems and help organizations make data-driven decisions.

Improving the performance of deep learning based speaker recognition systems – Apr 2018 to Present

Speaker recognition can be applied to a number of person authentication applications, including
credit card verification, over-the-phone secure access in call centres, and suspect identification by voice. Recent studies have found that the current state-of-the-art speaker verification systems, a significant amount of speech is required for development, enrolment and verification.

In this research, the performance degradation of deep learning speaker embedding based speaker verification system is studied when short-utterance data is used for speaker verification system development. Subsequently, a number of novel techniques will be proposed to improve the speaker verification performance in such conditions.


DIHARD Speech Diarization Challenge – Mar 2018 to May 2018

For DIHARD Speech Diarization Challenge, i-vector PLDA based speaker diarization system was developed using NIST 04, 05, 06 and 08 and SWB data and system results were submitted.


Forecasting of a solar photovoltaic power generation using deep learning approaches - Jan 2018 to Present

Nowadays power generation companies and individual people are moving to own solar photovoltaic cell based power generations. In order to build profitable power plant in a specific place, a number of factors needs to be investigated: (1) how the weather parameters vary, (2) how the solar irradiance vary, (3) what’s the correlation between weather parameters, and (4) forecasting solar irradiance and solar PV based power generation by analyzing weather parameters.

For this research, we have been awarded university research grant (LKR 800,000). In this research, deep learning techniques are proposed for short term forecasting.


Speakers in the Wild (SITW) 2015 speaker recognition challenge – Dec 2015 to Feb 2016

Participated Speakers in the Wild (SITW) 2015 speaker recognition challenge, and our system came 2nd as an organization, and our two systems came 3rd and 5th (out of 26 systems submitted).


NIST 2012 Speaker Recognition Evaluation – Jan 2012 to Dec 2012

For National Institute of Standards and Technology (NIST) Speaker Recognition Evaluation, we evaluated the state-of-the-art speaker recognition systems developed by ATVS and QUT research groups on the NIST SRE 2012 data. The project involved designing and developing algorithms for front-end processing, acoustic features, modelling, various back-end classifiers and score level fusion/calibration.


IP based Remote Monitoring System – Feb 2007 to July 2007

During our industry training period at Dialog broadband networks, we designed and developed an IP based remote monitoring system which is used to monitor the base station's AC power status, battery bank power level and temperature from NMC. The TCP/ IP stack and a mini Web server were implemented in the Ethernet/ microcontroller to configure the unit and auto transmit the sensor data from the remote station to NMC via IP packets. At the NMC, incoming data is processed and presented in web GUI which automatically alerts the required parties.




International Journals
6 results
2019
[6]A Study on Pairwise LDA for X-vector based Speaker Recognition (), In Electronics Letters, .
2018
[5]Improving the performance of GPLDA speaker verification using unsupervised inter-dataset variability compensation approaches (), In In International Journal of Speech Technology, .
[4]Improving PLDA Speaker Verification Performance using Domain Mismatch Compensation Techniques (, , , and ), In Computer Speech and Language, .
2017
[3]A Study on the Effects of Using Short Utterance Length Development Data in the Design of GPLDA Speaker Veri.cation Systems (, , , and ), In International Journal of Speech Technology, .
2014
[2]Improving Short Utterance I-vector Speaker Recognition using Utterance Variance Modelling and Compensation Techniques (, , , , and ), In Speech Communication, .
2013
[1]I-vector based Speaker Recognition Using Advanced Channel Compensation Techniques (, , , and ), In Computer Speech and Language, .
International Conference and Workshop Papers
27 results
2019
[27]LEAP Diarization System for the Second DIHARD Challenge (), In Proceed. of INTERSPEECH, International Speech Communication Association (ISCA), .
[26]A Study of X-vector Based Speaker Recognition on Short Utterances (), In Proceed. of INTERSPEECH, International Speech Communication Association (ISCA), .
2018
[25]Solar Irradiance Forecasting Using Deep Learning Approaches (), In National Undergraduate Research Symposium(Submitted), .
[24]Long-term Solar Irradiance Forecasting Approaches – A Comparative Study (), In International Conference on Information and Automation for Sustainability(Accepted), .
[23]Investigating Deep Neural Networks For Speaker Diarization In The Dihard Challenge (), In The IEEE Spoken Language Technology Workshop, .
[22]A Study on Solar PV Power Generation Influencing Parameters Using Captured Data from Faculty Of Engineering, University Of Jaffna Solar Measuring Station (), In In International Conference on Solar Energy Materials, Solar Cells & Solar Energy Applications, .
[21]A Study on Solar PV Power Generation Influencing Parameters Using Captured Data from Faculty Of Engineering, University Of Jaffna Solar Measuring Station (, and ), In International Conference on Solar Energy Materials, Solar Cells & Solar Energy Applications, .
2016
[20]SPEAKERS IN THE WILD (SITW): The QUT Speaker Recognition System (, , , , , and ), In 2016 International Speech Communication Association (ISCA), .
[19]Short Utterance Variance Modelling and Utterance Partitioning for PLDA Speaker Verification (, , , and ), In 2016 International Speech Communication Association (ISCA), .
[18]Domain adaptation based Speaker Recognition on Short Utterances (, , and ), In arXiv preprint arXiv:1610.02831, .
[17]DNN based Speaker Recognition on Short Utterances (, , and ), In arXiv preprint arXiv:1610.03190, .
[16]Improving Short Utterance PLDA Speaker Verification using SUV Modelling and Utterance Partitioning Approach (, , and ), In arXiv preprint arXiv:1610.04965, .
2015
[15]Improving out-domain PLDA speaker verification using unsupervised inter-dataset variability compensation approach (, and ), In 2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), . [doi]
[14]Improving PLDA Speaker Verification using WMFD and Linear-weighted Approaches in Limited Microphone Data Conditions (, and ), In Proceed. of INTERSPEECH, International Speech Communication Association (ISCA), .
[13]Dataset-Invariant Covariance Normalization for Out-domain PLDA Speaker Verification (, , and ), In 2015 Proceed. of INTERSPEECH, International Speech Communication Association (ISCA), .
[12]Investigating in-domain data requirements for PLDA training (, , and ), In 2015 International Speech Communication Association (ISCA), .
[11]The QUT-NOISE SRE Protocol for the Evaluation of Noisy Speaker Recognition (, , , and ), In 2015 International Speech Communication Association (ISCA), .
2014
[10]Improving PLDA speaker verification with limited development data (, and ), In 2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), . [doi]
[9]Short Utterance PLDA Speaker Verification using SN-WLDA and Variance Modelling Techniques (, and ), In 15th Australian International Conference on Speech Science and Technology, The Australian Speech Science & Technology Association, .
2013
[8]Improving Short Utterance based I-vector Speaker Recognition using Source and Utterance-Duration Normalization Techniques (, , , , and ), In Proceed. of INTERSPEECH, International Speech Communication Association (ISCA), .
[7]Improving the PLDA based Speaker Verification in Limited Microphone Data Conditions (, , , , and ), In Proceed. of INTERSPEECH, International Speech Communication Association (ISCA), .
2012
[6]Weighted LDA techniques for i-vector based speaker verification (, , , , and ), In 2012 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), . [doi]
[5]PLDA based Speaker Recognition on Short Utterances (, , and ), In Odyssey Speaker and Language Recognition Workshop, .
[4]PLDA based Speaker Verification with Weighted LDA Techniques (, , and ), In Odyssey Speaker and Language Recognition Workshop, .
[3]JFA based Speaker Recognition using Delta-Phase and MFCC features (, and ), In 14th Australian International Conference on Speech Science and Technology, The Australian Speech Science & Technology Association, .
[2]ATVS-QUT NIST SRE 2012 SYSTEM (, , , , , , and ), In Proceedings of NIST Speaker Recognition Evaluation 2012, .
2011
[1]i-vector based speaker recognition on short utterances (, , and ), In 2011 International Speech Communication Association (ISCA), .
Other Publications
1 result
2014
[1]Speaker Verification using I-vector Features (), PhD thesis, Queensland University of Technology, .

Sep 2016 to Present

Senior Lecturer & Coordinator for Renewable Energy Park
Department of Electrical and Electronic Engineering
Faculty of Engineering,
University of Jaffna.
Teaching:
 Applied Electricity, Signals & Systems, Control Systems, Advanced Digital Signal Processing.
Research:
 Improving the performance of speaker embedding based speaker recognition systems,
 Solar PV power forecasting using deep learning approaches.
Administrative work:
 Coordinator for renewable energy park,
 Established renewable energy park research center.


Feb 2018 to Apr 2018

Visiting Researcher
Speech, audio, image and video technology (SAIVT) lab
Queensland University of Technology, Brisbane.
Research:
 Participated DIHARD speaker diarization challenge,
 Investigated domain adaptation approaches to improve the performance of speaker embedding based speaker recognition systems.


Oct 2017 to Jan 2018

Visiting Lecturer
Faculty of Applied Science, South Eastern University
Teaching:
 Data acquisition & signal processing.


Dec 2017 to Apr 2018

Visiting Lecturer
B.Ed in Engineering Technology Program
Teaching:
 Principle of electronics.

Nov 2014 to Sep 2016

Postdoc Researcher
Speech, audio, image and video technology (SAIVT) lab
Queensland University of Technology, Brisbane.
Research:
 This work implemented and examined the machine learning algorithms to improve the performance of speaker verification system in dataset mismatch and noisy conditions.
 Collaborated with other researchers and published papers in prestigious conferences and journals.
 Participated Speakers in the Wild (SITW) 2015 speaker recognition challenge, and our system came 2nd as an organisation, and our two systems came 3rd and 5th (out of 26 systems submitted).


Jan 2016 to Mar 2016

Visiting Lecturer
Department of Electrical and Electronic Engineering
Faculty of Engineering,
University of Jaffna.
Teaching:
 Signals & Systems.


Nov 2012 to Apr 2013

Visiting Researcher
Biometric Recognition Group ATVS
Autonomous University of Madrid, Madrid, Spain.
Research:
 Research conducted under the supervision of Prof. Joaquin Gonzalez-Rodriguez on Bayesian biometrics for forensics (BBfor2) project,
 The aim of that work was to achieve a reduction of the effect of utterance durations and achieving a significant reduction of the development data, which are key requirements for effective deployment of the speaker verification technology in forensic scenario.


Apr 2010 to Oct 2014

PhD Researcher
Speech, audio, image and video technology (SAIVT) lab
Queensland University of Technology, Brisbane.
Research:
 The research primarily conducted in the field of speaker verification. These studies examined machine learning techniques to improve current technologies,
 Implemented Gaussian mixture modelling, joint factor analysis, i-vector and probabilistic linear discriminant analysis based speaker verification systems using python and C++,
 Introduced dimensionality reduction techniques to compensate the channel variation and improve the performance of i-vector speaker verification systems,
 Introduced novel short utterance variance normalisation and modelling techniques to compensate the session and utterance variations and improve the performance of short utterance i-vector speaker verification systems,
 Participated National Institute of Standards and Technology (NIST) 2012 speaker recognition evaluations.

Sep 2008 to Apr 2010

Business Development & Technical Support Engineer
Sri Lanka Telecom PLC, Colombo, Sri Lanka.
Responsibilities:
 Applied business analytical skills to increase sales, reduce churn and decrease operational costs.
 Analyzed telecommunication market and proposed transmission ring network based on customer’s requirement.
 Negotiated with international telecommunication service providers for buying international transmission links and internet ports.
 Prepared service-level agreement (SLA) for transmissions link and data services.
 Conducted training session to account managers.


Feb 2007 to Jul 2007

Trainee Engineer
Dialog Broadband Networks (Pvt) LtD, Colombo, Srilanka.