PyTorch 0.4.0, Google Brain Tokyo, QuickNLP, Multilingual NLU, PeerRead dataset, PyTorch GAN, ML Openness, SGD Earth, DL for Alzheimer’s Detection,…

elvis
DAIR.AI
Published in
4 min readApr 30, 2018

Welcome to the 12th Issue of the NLP Newsletter! Here is this week’s notable NLP news!

On People…
A statement has been circulating for ML people to sign against “closed access or author-fee publication” (We definitely signed! Let’s keep ML research open) — Link

Are emojis ruining the way young people use the English language to communicate? — Link

Hands down one of the best AI essays of the year by Professor Michael Jordan (Artificial Intelligence — The Revolution Hasn’t Happened Yet) — Link

Multilingual natural language understanding by Spacy (Video) — Link

Why technical experts need to get better at telling stories — Link

Bias in word embeddings? (Google Research Blog) — Link

On Education and Research…
Paper revisiting on how to make better choices with batch training in neural networks — Link

Paper discussing when and why word embeddings are good for neural machine translation — Link

Olive Oil is Made of Olives, Baby Oil is Made for Babies (Paper Summary) — Link

Paper discusses how to go about conducting evaluation on “Deep Semi-Supervised Learning Algorithms” — Link

Using deep learning to detect linguistic cues of Alzheimer’s disease patients — Link

Summary of interesting NLP Papers and Research (Fast and easy reads!) — Link

How Natural Language Inference Models “Game” the Task of Learning — Link

On Code and Data…
PyTorch 0.4.0 is released (Trade-off memory for compute, Windows support, 24 distributions with cdf, variance etc., dtypes, zero-dimensional Tensors, Tensor-Variable merge, faster distributed, perf and bug fixes, CuDNN 7.1) — Link

Quick NLP is a deep learning nlp library inspired by the fast.ai library — Link

“PeerRead is a dataset of scientific peer reviews available to help researchers study this important artifact” — Link

Executing gradient descent on Earth — Link

GANs in 50 lines of code with PyTorch — Link

A guide to conducting sequence prediction (one of the hottest trends in deep learning) with Python — Link

Alphabetical list of NLP datasets — Link

On Industry…
Google expands Google Brain team in Tokyo (now accepting applications) — Link

OpenAI holds transfer learning contest using Sonic The Hedgehog game — Link

Very nice video tutorial by Uber AI Labs on “Measuring the Intrinsic Dimension of Objective Landscapes” — Link

How NLP offers a bright future for airlines and passengers — Link

Scientists plan huge European AI hub to compete with US — Link

Quote of the Week…
“We see no role for closed access or author-fee publication in the future of machine learning research and believe the adoption of this new journal as an outlet of record for the machine learning community would be a retrograde step.” (@tdietterich)

Visualization of the Week…

“Sorry officer, my car has old version of TensorFlow, I’ll update it tonight” — Link

Worthy Mentions…
Implementing YOLO v3 with PyTorch — Link

NLP News by Sebastian Ruder (Issue 21) — Link

Speakeasy — Fun article explaining how aliens can help our digital assistants with linguistic assistance (sounds mostly fictional but it’s actually very insightful) — Link

A nice summarized version of the MIT Probability course — Link

Paper on the Theory of Mind — Link

Why deep learning is perfect for NLP — Link

5 Fantastic Practical Natural Language Processing Resources — Link

Our previous newsletter (Issue 11) — Link

Message from the Editor
Hello everyone! Thanks for the massive support you have given to this newsletter. I realized the tremendous good it can do for experts in the field and those who are beginning. The previous newsletter did so well that I got excited to see how much people were interested in such newsletter (3K+ views) 👏. I hope to keep bringing more of the best of NLP and ML news in the weeks that follow. Also, come over to @omarsar0 and say hi!

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.

--

--