In the News
McKinsey report on AI
- Tech companies like Baidu and Google spent between $20B to $30B on AI in 2016. 90% was spent on R&D and 10% on AI-related acquisitions.
- 66% of all AI investments in 2016 were to US-based companies. China was second with 17%.
Full report available here
Microsoft announces new research unit to take on Deepmind
The new unit will focus on General Intelligence along with issues in health, environment and education.
Learning
Apple Machine Learning Journal
Apple launched a Machine Learning Journal to share some of the work carried out by its researchers. It's interesting to see such a big change for an otherwise secretive Apple. The first post is on "Improving the Realism of Synthetic Images".
Revisiting the Unreasonable Effectiveness of Data
Is larger always better? This study explores the impact of XL image sets (300M images) and shows that there is indeed a benefit — growing linearly with the log of the dataset size.
The Bias-Variance Dilemma
There is a delicate balance to be found for ML models between concept generalization and fine-grained representation of data.
Software tools & code
Interpreting neurons in an LSTM network
Interesting post that takes a look at what individual neurons in an LSTM networks actually learn when performing transliteration tasks.
Neural coreference resolution for chatbots
Open-source state-of-the-art coreference resolution library in Python. Uses word embeddings and neural networks to link together mentions of entities in raw text or dialogue. Comes with a nice online demo.
Using Deep Learning to Create Professional-Level Photographs
Panorama selection from Street View images, followed by automated image improvements to produce professional-looking photos.
Hardware
New AWS GPU instances
Amazon Web Services introduced their new G3 instance type, powered by NVIDIA Tesla M60 GPUs.
About
This newsletter is a collection of AI news and resources curated by @dlissmyr. If you find it worthwhile, please forward to your friends and colleagues, or share on your favorite network!
Share on Twitter · Share on Linkedin · Share on Google+
Suggestions or comments are more than welcome, just reply to this email.
Thanks!