In this first NLP paper reading session we will discuss the paper
"A Unified Architecture for Natural Language Processing: Deep Neural Networks
with Multitask Learning"
which won the test-of-time award in the 2018 International Conference of Machine Learning. This paper was written by Ronan Collobert and Jason Weston, and published in 2008 in the same conference.
We will talk about concepts ranging from word embeddings to training auxiliary tasks as means to improve modeling power, and understand why some of the concepts presented ten years ago are still widely-used today.
January 26 (Sat): Reading Guidelines are published.
January 26 - February 1: Self reading, and preparing.
February 2: Paper reading session.
Paper Reading Schedule:
12:50 (Optional) Meet for lunch.
14:00 - 14:20 Presentation of the paper by Jorge Balazs.
14:20 - ~16:00 Paper discussion.
The Reading Guidelines will include important points to consider while reading the paper, as well as related literature and concepts. It will also serve as guide during the discussion.
Non-experts are also welcome. The only requirement is to read the paper and follow the reading guidelines.
Reading Guidelines: https://discuss.mltokyo.ai/t/paper-reading-guidelines/241
Further Instructions and discussion: https://discuss.mltokyo.ai/t/nlp-pr-1-a-unified-architecture-for-natural-language-processing-deep-neural-networks-with-multitask-learning/242
-- ENGINEERS AND RESEARCHERS ONLY. NO RECRUITERS, SORRY. --
Join us on:
[Discourse] For MLT workshop materials: http://discuss.mltokyo.ai