Polysemy Embeddings, Semi Adversarial Networks, ULMFiT, Color Naming, Sentiment Style Transfer,…
Welcome to the 15th Issue of the NLP Newsletter! Here is this week’s notable NLP news!
On People…
Judea Pearl, AI pioneer, gives his advice on how we should move AI forward — Link
Meet the new Kaggle grand-master — Link
Color naming shaped by communicative need — Link
A simple approach to sentiment and style transfer — Link
Can a CNN learn when to reject and accept papers? — Link
On Education and Research…
Introducing ULMFiT a universal language model for fine-tuning through transfer learning (useful for many NLP tasks) — Link
Word Embeddings to Polysemy Embedding — Link
This paper introduces a la carte embedding, a simple and general alternative to the usual word2vec-based approaches for building word embeddings — Link
What can be learned from extrapolating to examples outside the training space — Link
Modeling semantics with graph neural networks, which help to build a knowledge-based Q/A system — Link
Deep Learning Winter course by Andrej Karpathy — Link
On Code and Data…
PyTorch implementation of a Semi Adversarial Network — Link
Computing derivatives with PyTorch — Link
Part of Speech tagging with LSTMs — Link
On Industry…
I doubt this is possible 👐, nevertheless, here is DeepMind’s shot at understanding how the brain thinks — Link
Should we really be frightened about futuristic AI — Link
How Google researchers plan to advance the study of Semantic Textual Similarity — Link
Quotes of the day…
A potential useful way to detect spelling errors using word embeddings — Link
Interesting point of view by Denny Britz — Link
Illustrations of the day…
Learn more about BLEU and meaning representation in this upcoming new paper — Link
Worthy Mentions…
Demystifying Generative Adversarial Networks — Link
Solving detection of fake news through AI and NLP — Link
Universal language model to boost your NLP models — Link
NLP Newsletter by Sebastian Ruder — Link
If you spot any errors or inaccuracies in this newsletter please open an issue. Submit a pull request if you would like to add important NLP news here.