Member of the digital campaign team @ La République En Marche ! - Montréal
I am a PhD student who has been working on reinforcement learning for adaptive spoken dialogue systems. I have a computer science and artificial intelligence background. Throughout my Phd and previous internships, I have gained expertise on Bayesian frameworks, reinforcement learning and supervised and unsupervised learning. I have also practiced data visualization with R. I have software
I am a PhD student who has been working on reinforcement learning for adaptive spoken dialogue systems. I have a computer science and artificial intelligence background. Throughout my Phd and previous internships, I have gained expertise on Bayesian frameworks, reinforcement learning and supervised and unsupervised learning. I have also practiced data visualization with R. I have software implementation experience in Java and C++.
PHD Student @ I prposed algorithms to learn, from examples of dialogues, the reinforcement learning parameters of an adaptive spoken dialogue system (see description below and list of publications). From October 2011 to Present (4 years 3 months) Researcher PHD Student @ Among the NAtural DIAlogue (NADIA) team, I worked on machine learning for adaptive spoken dialogue systems.
The main tasks were the following:
- Java implementation of algorithms that learn from dialogues the parameters (reward function and state space representation) of a reinforcement learning-based spoken dialogue system.
- Improvement of the dialogue manager of a spoken dialogue system for appointment scheduling.
- Test of the dialogue strategies with Orange Labs volunteering employees. Users were asked to fill in a questionnaire after each dialogue.
- Definition and extraction of dialogue features from logs.
- Supervised learning applied to automatically estimating user satisfaction based on the chosen feature set.
- Feature selection for reinforcement learning. From October 2011 to October 2014 (3 years 1 month) Internship @ The purpose was to study the convergence properties of a reinforcement learning algorithm embedded in a spoken dialogue system.
For this task :
- I implemented in Java a dialogue simulator by parsing XML files describing the behavior of a spoken dialogue system and by inferring user behavior from real dialogues with the system.
- I proposed an algorithm to estimate the convergence time of reinforcement learning.
The result of this work was published in the proceedings of SIGDIAL 2013. From February 2011 to July 2011 (6 months) Summer Internship @ The object of this internship was to endue a dyslexia training software with adaptive capabilities. The learning environment was modeled with a Bayesian probabilistic framework and the software adapted the level of difficulty of the exercises proposed to the user according to the user's performance. To test the performance of the algorithm, I implemented simulations in C++. From July 2010 to September 2010 (3 months) Mathematics Tutor @ Mathematics lessons to a junior high school student. From 2010 to 2010 (less than a year) Grenoble Area, France
Doctor of Philosophy (PhD), Computer Science @ Université de Lorraine From 2011 to 2015 Coursera From 2014 to 2014 Machine Learning Summer School 2014Master's degree, Artificial Intelligence and the Web @ Master of Science in Informatics at Grenoble (MoSIG) From 2010 to 2011 Engineering Degree, Computer Science and Applied Mathematics @ Ecole Nationale Supérieure d'Informatique et de Mathématiques Appliquées de Grenoble From 2008 to 2011 Layla Asri is skilled in: C++, Algorithms, Computer Science, Artificial Intelligence, Machine Learning, Java, C, Linux, SQL, R, Software Engineering, Eclipse, Unix, LaTeX, Git, Python
Looking for a different
Get an email address for anyone on LinkedIn with the ContactOut Chrome extension