RALI/DIRO

About Me

Philippe Langlais

I am Full professor at DIRO (Computer Science and Operational Research of University of Montreal) working on Natural Language Processing at RALI. I did my Phd at University of Avignon (on the use of prosody in Speech Recognition) and my master at University of Aix-Marseille (on a music dictation system). I also visited IDIAP/Martigny for some years (collecting a large speech database), KTH/Stockholm (working on Speech Technology for 2nd Language Acquisition), and LIMSI/Paris-Orsay (now LISN).

Research Topics.

I am interested in many NLP related topics, including core-technology components (e.g. named-entity recognition, information extraction), and specific applications (e.g. Question-Answering, Machine Translation, Legal Prediction, Bilingual Concordancer). I have written a number of papers, most of which can be found here.

Interests: multilingual applications (machine translation, sentence and word alignment), information extraction, question-answering, data augmentation, data preparation, bias in datasets and models

To students

I am always looking for motivated students. This means students that enjoy reading scientific articles, implementing things, and that care about data. Outperforming current models on specific benchmarks is not to me an end, but more a means to understand limitations of current state-of-the-art.

I am receiving a lot of spontaneous applications from abroad students, and it is important to understand that tuition fees are high and that grants to cover those fees are hard to get, not even mentioning immigration issues.

To professionals

I am collaborating with several companies. My philosophy in such collaborations is that students should develop academical skills first, considering professional settings (datasets, applications) as a challenge. Not all companies are ready to welcome students in such an open environment but here are two that does: LexRockAI and Huawei.

About us
PhD Students

Current

  • Frédéric Piedboeuf Data augmentation, Generative models
  • Olivier Salaün Legal Justice Prediction
  • Peng Lu Efficient training, Quantum Learning
  • Antoine Elie Biases in Question-Answering
  • David Romain Djoumbissie Financial Prediction

MSc Students

Current

  • Juan Felipe Durand Unsupervised fine-tuning of Question-Answering
  • Nathan Migeon Knowledge in Question-Answering
  • Sanjeev Pratti Controling Machine Translation with GANs
  • Yves Blain-Montesano Analogical Learning
About us

What my friends say

Client Image
Chilly

I wish I had Phil as a CS prof instead of being taught piano at McGill.

Client Image
Jérôme Minière

Phil must be a very good prof since I coached him how to prepare a powerpoint.

Client Image
Tigran Hamasyam

Phil is the best professor ever. He taught me computer science, but I eventually became known as a pianist.

Adapted from © DevCRUD