Bo Li

Speech Scientist, Cerence Inc.

Email: bolimcmaster@gmail.com


Research

Current Research topic

      • Removing gender bias from pre-trained word-embedding.

      • Sentence modeling for downward NLP task.

Google Scholar

https://scholar.google.ca/citations?user=eRC_qY8AAAAJ&hl=zh-CN

Skills

C++, C, C#, Matlab, Python, Java, scikit-learn, tensorflow.

Education

      • P.h.d: Control Science and Engineering, Harbin Institute of Technology, P. R. China, 2014.

      • M.A.Sc.: Electrical and Computer Engineering, McMaster University, Canada, 2013.

      • Master: Communication and Information Engineering, Harbin Institute of Technology, P. R. China, 2010.

      • B.S.: Electronic and Information Engineering, Harbin Institute of Technology, P. R. China, 2008.

Professional Experiences

Post-doctorate Fellow, Electrical Engineering, École de Technologie Supérieure, June 2016 - July 2017

        • Led research on a machine learning project. Feature selection algorithms were evaluated for feature selection. Supervised learning algorithms, such as linear discriminate analysis, support vector machine, were utilized for training and prediction.

Research Engineer, Electrical and Computer Engineering, McMaster University, January, 2015 - December, 2015

        • Research on projection matrix design using prior information in compressive sensing. A majorization minimization based algorithm was proposed to design the projection matrix with prior information on the sparse signal. The proposed algorithm was evaluated to be effective theoretically and experimentally.

        • Research on projection matrix design in adaptive compressive sensing. A water-filling based projection matrix design algorithm was proposed to design the projection matrix in the second stage of the two-stage adaptive compressive sensing. The designed projection matrix focus the sensing energy on the positions of the sparse signal which are of non-zero magnitude. With the designed projection matrix, the SNR of the measurement signal improves and the mean squared error of the estimated sparse signal is reduced.

Publications

Journal Articles

Conference Papers with Proceedings

1. Bo Li, Wassim Bouachir, Rafik Gouiaa and Rita Noumeir. Real-time recognition of suicidal behavior using and RGB-D camera, International conference on image processing theory, tools and applications, 2017.

2. Jia Li, Qiang Wang, Yi Shen and Bo Li, performance analysis of OMP algorithm in recovering magnitude decaying sparse signals in compressive sensing, 2013 IEEE Conference on I2MTC, pp. 1781-1785, 2013.

3. Yi Shen, Bo Li and Zhenghua Wu, fast OOMP algorithm and its application in compressed sensing. 2011 IEEE Conference on I2MTC, pp. 520-524, 2011.

Papers under Review

Papers in preparation

Awards & Honors

      • Nominee of GSA Keith Leppmann Teaching Assistant Excellence Award ( McMaster University ), 2013

      • International Excellent Student Scholarship ( McMaster University ), 2012

      • Graduate Research Scholarship ( McMaster University ), 2011

      • First class Postgraduate Scholarship ( top 5 percents, Harbin Institute of Tech. ), 2009

      • Provincial Excellent Graduate Award ( 2/130, Harbin Institute of Tech. ), 2008

      • Excellent Bachelor Thesis Award ( 2/130, Harbin Institute of Tech. ), 2008

      • First Class Public Scholarship ( top 3 percents, Harbin Institute of Tech. ), 2006

Notes

Bias, variance and k-fold cross validation.

Connections