TensorboardX, NLP Best Practices, Cirq, Subjectivity in AI, MojiTalk, Seq-to-Seq Debugging, Data Portability,…
Welcome to the 22nd Issue of the NLP Newsletter! Here is this week’s notable NLP news! Tensorboard for PyTorch, NLP best practices, Subjectivity in AI, Portability of data, Google’s new RNN, Major issues in NMT, and much more.
On People…
Find out how Google wants to enable the portability of data, which involves giving users full control of their data — Link
An online textbook discussing fairness in machine learning — Link
Should subjectivity be allowed to enter the discourse in the science of the artificial (artificial intelligence and artificial life) — Link
Will NLP change how we communicate in the future? — Link
Companies developing AI sign agreement at IJCAI pledging to avoid development of lethal autonomous weapons — Link
Find out how templates and other forms for documenting patients data can undermine big-data efforts — Link
On Education and Research…
Prodigy releases best practices for NLP projects, which include discussions and guides on how to better carry out your NLP research projects — Link
Delip Rao discusses six major issues in neural machine translation (he also addresses the issues with irresponsible AI reporting) — Link
Check out all the talks given at NAACL 2018 (videos) — Link
Rachel Thomas provide a nice introduction to neural architecture search and why this sub-field has been considered important in making machine learning more accessible — Link
A new study provides suggestions on how to ensure that your company’s AI projects succeed — Link
Learn how to generate neural text using conditional language models — Link
An in-depth analysis of how deep learning is used for sound classification — Link
On Code and Data…
With TensorboardX, now you can use Tensorboard with many other deep learning frameworks such as PyTorch, Chainer, MXNet, among others — Link
Reproducible machine learning with PyTorch and Quilt data registry — Link
Code release for MojiTalk which is a system used to generate emotional responses at scale — Link
Listen to a beginner discussing his experience in using tensors with Tensorflow — Link
On Industry…
Google AI research team releases new paper on how an improved version of RNN can be used to improve the accuracy of automatic interpretation of connectomics data — Link
A new study reports that AI will “contribute around $15.7 trillion to the global economy by 2030” — Link
IBM releases a tool for visually debugging sequence-to-sequence models — Link
Google releases Cirq, an open source tool which enables researchers to write quantum algorithms for quantum processors such as those used for NISQ computers — Link
How AI and NLP can enhance human communication — Link
Find out why the last Sir co-founder has retired from Apple — Link
Quotes of the day…
Worthy Mentions…
A visual guide showing how Decision Trees work — Link
A fun beginner tutorial for NLP — Link
Hal Daume shares his thoughts on why diversity of speakers in conferences matter — Link
How do you transition from being a data analyst to data scientist? — Link
If you spot any errors or inaccuracies in this newsletter please comment below.
