Editorial
Thoughts From Your Humble Curators
Happy 2018, everyone! We start off this year by learning China is building a $2.1 billion AI park, would China AI’s development catch up with U.S.? Let’s find out in our News section.
We also have interesting contents in our Blogs section. Two gems we recommend is Denny Britz’s “AI and Deep Learning in 2017 – A Year in Review” and Prof. Rodney Brooks dated prediction. The former is a comprehensive review of research development of AI/DL in 2017, the latter is a thoughtful prediction on the future of SDC and DL by Prof. Brooks.
As always, if you like our newsletter, feel free to subscribe and forward it your colleagues/friends!
The AIDL Page
We recently created a facebook Page, which compliments our Groups. You can think of it as your one-stop shop for AI/DL-related news.
So what’s the difference between the Page and the Group then? It’s simple: Page is more suitable for posting news, whereas Group is better for discussion, hang out and chill. Also as a routine, AIDL Page would post around 3-5 articles daily on the AIDL group, those are posts chosen from large amount of candidate posts (~10-30). So you may also see AIDL Page as your AI news filter.
If you have time, visit our Page and give us a nice 5-star review? Thanks!
Sponsor
Video conferencing that just works in your browser. No download. No PIN code. No login
Just launch it from Cyclop.io, or add to your Slack team and launch from Slack. Works in your browser. No Download. No login. Get up and running in seconds. Multiply team’s productivity
News
China is Building A $2.1 Billion AI Park
China is going to build an AI Park which can host 400 businesses, with focus on AI and deep learning. This seems to be in line with China’s latest Five Year Plan. It’s well known to us that China has a strong AI presence. For example, we know that Resnet was first created by Chinese researcher in MSRA, and recent Imagenet competition is dominated by Chinese companies/universities.
Many would naturally ask – what makes U.S. in the AI competition? We believe that U.S. still has advantage that includes the presence of many AI powerhouses such as Google, Nvidia, and academic research institutes such as CMU, Stanford and MIT. U.S. is also the first country which picked up deep learning research to applications. So it would take a while for other countries to catch up.
Before we go, in an interview early this early at the Late Show with Stephen Colbert. Colbert asked Neil Degrasse Tyson on why the current largest telescope is not run by U.S. but by China. He said “We lost our mojo. […] The Science mojo is this has never been done before, we gonna invest in it, we gonna first at it. If you don’t have those three pistons inline, other people will do it!” Would that happen in US A.I. research as well? Hopefully the US government would create enough incentives to keep such “AI mojo”.
Blog Posts
Supercomputer from 21k Voltas
IBM is hooking up 21k Voltas V100 and 3.500 AC922 together and form a supercomputer. The link will bring you to an impressive video of how the system was built.
Analysis of ML/NLP publication in 2017 by Mark Rei
This is an analysis by Mark Rei on different publications in ML/NLP. Perhaps what really caught our eyes in CMU was in the front-place for several categories. But as star researchers such as Chris Dyer is leaving for industry research sites such as DeepMind, would the ranking changes?
AI and Deep Learning in 2017 – A Year in Review
Denny Britz wrote one of the best 2017 reviews on deep learning and AI. Unlike most, Britz’ approach is comprehensive and you should just bookmark it to look up useful links/resources.
Prof. Lecun on Cargo Cult AI
Prof. Yann LeCun on a recent TechInsider’s interview with Sohpia:
“This is to AI as prestidigitation is to real magic.
Perhaps we should call this “Cargo Cult AI” or “Potemkin AI” or “Wizard-of-Oz AI”. In other words, it’s complete bullsh*t (pardon my French). Tech Insider: you are complicit in this scam.”
Some criticize AIDL members more negative attitude towards Sophia, we just hope to point out that we are not alone.
Rodney Brooks’ Dated Predictions
Rodney Brooks thoughtful and dated prediction of self-driving cars and deep learning. He is not as famous as Musk or as prominent as physicists. But his predictions look way more realistic.
Open Source
Deep Learning Model Converters
Here is a great link on how to convert models from different deep learning frameworks. There are multiple converters which comes in handy.
About Us
You may donate to this link to help our operations.
Join our community for real-time discussions with this iOS app here: https://itunes.apple.com/us/app/expertify/id969850760