Deep Language Workshop 2017

12 September 2017, Université de Montréal

Deep Learning has become a key ingredient of progress towards computers which understand and generate language, following pioneering work at MILA more than 15 years ago on neural language models. Modelling, understanding and generating language is a crucial component of artificial intelligence and progress in this field and is also likely to be economically transformative.

This one-day MILA workshop is an opportunity to share advances in this area and is being sponsored by two MILA partners, Microsoft (in particular Maluuba) and Druide.

Attendance is by invitation.

Program

 Time Speaker Title
09:00 Yi-Min Wang and Harry Shum Opening and Welcome
09:15 Sam Pasupalak and Kaheer Suleman Maluuba: Building Literate Machines
09:25 Adam Trischler Machine Reading Comprehension
09:45 Harm van Seijen Achieving Above-Human Performance on Ms. Pac-Man by Reward Decomposition
10:05 Simon Lacoste-Julien SeaRNN: Training RNNs with Global-Local Losses
10:25 COFFEE BREAK
10:40 Jianfeng Gao Recent Progress on MRC and Dialogue
11:00 Lihong Li Recent Progress on Reinforcement Learning
11:20 Rangan Majumder Machine Comprehension at Bing
11:40 Weizhu Chen Conversational AI for Enterprises and its Research Challenges
12:00 LUNCH BREAK
13:30 Yoshua Bengio Towards Deep Language Understanding
14:05 Alexandre de Brébisson RNNs for Speech Synthesis
14:25 Rosemary Ke Hierarchical attention for QA
14:45 COFFEE BREAK
15:00 Dmitry Bahdanau Natural Language Understanding Challenges
15:20 Kian Kenyon-Dean Event Coreference Resolution
15:40 Ethan Perez Learning Visual Reasoning Without Strong Priors
16:00 – 16:10 Wrap-up

Location

The workshop will be held on September 12 2017 at Université de Montréal, in room M-415 of Roger-Gaudry building. Wing M is located to the right of the main entrance coming in from Cour d’honneur (L-4).

Address: 2900 Boulevard Edouard-Montpetit, Montréal, QC H3T 1J4

     

 Sponsors

Organization

Organizer: Yoshua Bengio