29 January - 4 February 2018, Moscow
DeepHack.Babel
Congratulations for the winner of our hackathon
Lost in Translation
Lost in Translation
What is DeepHack?
А 24/7 week-long sci-hack school with the goal to solve fundamental problems of AI
Topics we cover
- Neural Networks
- Deep Learning
- Machine Translation
- Deep Learning Algorithms
The main goal of hackathon is semi-supervised Machine Translation: Machine Translation which makes use not only from (often scarce) parallel data, but also from (potentially unlimited) untranslated texts in source and target languages. The participants will need to train an MT model on a very small parallel corpus and boost its performance using monolingual non-parallel data.
Confirmed speakers
- Kyunghyun Cho
- Andre Martins
- Lucia Specia
- Maja Popovic
- Qun Liu
- Ruslan Salakhutdinov
- Serge Sharoff
- Yulia Tsvetkov
- Maxim Khalilov
- Romain Paulus
- Illia Polosukhin
- Maksim Kretov
- Eduard Klyshinsky
- Кonstantin Anisimovich
Poster session
School program
Monday, January 29
09:00–10:00
Registration of participants and checking into the dormitory
10:00–11:00
Welcome speech by organizers
11:00–11:30
Kickoff meeting and team building
Maja Popovic
Automatic evaluation and error classification of machine translation output
Tuesday, January 30
Wednesday, January 31
Thursday, February 1
Serge Sharoff
Comparable corpora as cure-all remedy: possibilities for mapping languages without parallel resources
Konstantin Anisimovich
20:30–22:00
Использование параллельных текстов для переноса синтаксической модели на новый язык
Friday, February 2
Kyunghyung Cho
Continuous representation of language and its implications - In the case of neural machine translation
Saturday, February 3
Drafting of posters and presentations of solutions
9:00–15:00
Sunday, February 4
Olga Gureenkova
DeepPavlov 0.0.1 Hard Adventures
An open source library for building end-to-end dialog systems and training chatbots.
14:00–15:00
Poster session
Announcement of results and awarding the hackathon winners
Closing reception
Our News
Attention Is All You Need
Участники обучали систему машинного перевода на непараллельных данных, то есть на примерах, для которых нет перевода на другой язык. Рассказываем, зачем это нужно.
Научная школа - серия лекций от ведущих мировых специалистов по машинному переводу.
The hackathon will focus on machine translation, with a special emphasis on neural machine translation (NMT)
Цель хакатона, который пройдёт в МФТИ, разработать методы обучения машинного переводчика без обучающих примеров
Систему машинного перевода нужно обучать на непараллельных данных — то есть без привлечения учителя.
Organizer
Neural Networks and Deep Learning Lab
Partners
Info partners
Previous DeepHack hackathons:
2017 DeepHack. Turing
2017 DeepHack.RL
2016 DeepHack.Q&A
2015 DeepHack.Game
Looking forward for next #deephack hackathons