Ludovic Trottier

Ph.D. Candidate
Department of Computer Science and Software Engineering
Laval University

I am a Ph.D candidate at Laval University under the supervision of Pr. Brahim Chaib-draa and Pr. Philippe Giguère. Here is my resume.

Research Interests

  • Machine learning: deep learning, dictionary learning
  • Speech Recognition: acoustic modelling, feature representation

Ludovic Trottier, Philippe Giguère, Brahim Chaib-draa: Parametric Exponential Linear Unit for Deep Convolutional Neural Networks. arXiv preprint arXiv:1605.09332   pdf

   Ludovic Trottier, Philippe Giguère, Brahim Chaib-draa: Dictionary Learning for Robotic Grasp Recognition and Detection. arXiv preprint arXiv:1606.00538
 Ludovic Trottier, Brahim Chaib-draaPhilippe Giguère: Incrementally Built Dictionary Learning for Sparse Representation. 2015 International Conference on Neural Information Precessing (ICONIP).

Ludovic Trottier, Brahim Chaib-draaPhilippe GiguèreFeature Selection for Robust Automatic Speech Recognition: A Temporal Offset Approach. International Journal of Speech Technology (IJST) 2015.

 Ludovic Trottier, Brahim Chaib-draaPhilippe GiguèreTemporal Feature Selection for Noisy Speech Recognition. Advances in Artificial Intelligence, Springer International Publishing, 2015: 155-166

Ludovic Trottier, Brahim Chaib-draa, Philippe GiguèreEffects of Frequency-Based Inter-frame Dependencies on Automatic Speech Recognition. Canadian Conference on AI 2014: 357-362



 August 7 2016
Parametric Exponential Linear Unit
 April 15 2016
Recent Advances in Convolutional Neural Networks
 IGGG pdf
 November 9 2015
 Incrementally Built Dictionary Learning
for Sparse Representation
 July 23 2015
Deep Learning for Detecting Robotic Grasps
IGGG pdf
 June 29 2015
Unsupervised Feature Learning Based on Clustering
 June 4 2015
Temporal Feature Selection for Noisy Speech Recognition
 AI 2015
November 11 2014 Spike-and-Slab Sparse Coding  DAMAS
October 7 2014  Probabilistic Interpretation of Structured Sparse Coding  DAMAS   
May 8 2014  Effects of Frequency-Based Inter-frame Dependencies on Automatic Speech Recognition AI 2014 pdf 
 March 18 2014 Sparse Coding for Speech Representation DAMAS  
 December 13 2013 Inter-frame Dependence Arising from
  Preceding and Succeeding Frames 
 July 30 2013 Introduction to Automatic Speech Recognition  DAMAS pdf 
 June 28 2013 Gibbs Sampling Acoustic Modelling for
Isolated Digits Recognition 
 June 6 2013  Reconnaissance automatique de la parole REPARTI  
 May 8 2013 Bayesian Inference for Hidden Markov Models DAMAS  
March 15 2013  Fast and Flexible Kullback-Leibler Divergence Based on Acoustic Modeling for Non-Native Speech Recognition DAMAS  
 December 21 2012 Solving SLAM Problem Using the Bayes Tree data structure DAMAS  
 July 25 2012 Off-Policy Actor Critic  DAMAS  
 June 13 2012 A Tutorial on Reinforcement Learning  DAMAS pdf 
 April 19 2012  Content-Based Spam Filtering DAMAS  
 June 10 2011 Gaussian Process  DAMAS  


Spam filter using information gain, inverse document frequency and chi squared to extract reliable features from text documents. The recognition was done using SVM, Naive Bayes and Hidden Markov Model.

Project for the course Concepts avancés pour systèmes intelligents, Pr. Brahim Chaib-draa, Laval University


Hand gesture recognition software for the control of pdf presentations without the usage of the mouse nor the keyboard. Histogram of gradients and SVM were used together in order to make this possible.

Project for the course Vision numérique, Pr. Denis Laurendeau, Laval University

  Simultaneous Localization and Mapping (SLAM) is the problem where a robot has to construct a map of its environment and to locate itself in it. In this project, we looked at the limitations of QR and Cholesky factorisation when trying to solve the smoothing problem using the optimisation framework.

Project for the course Introduction à la robotique mobile, Pr. Philippe Giguère, Laval University.


Object tracking is the problem of creating a robot that can follow a moving object. In this project, we tried to simulate skiptracing from a robot point of view : both the skiptracer and the fugitive were robots. We tested reinforcement learning and bynamic bayesian networks.

Project for the course Introduction à la recherche, Pr. Brahim Chaib-draa, Laval University.