U.S. widens trade blacklist to include some of China’s top AI startups

Re:U.S. widens trade blacklist to include some of China’s to... by daniel_iversen on Wednesday October 09, 2019 @02:41AM (#21200684)

Apparently China is already leading the world as far as AI goes (I first saw this in Mary Meeker’s internet report this year I think it was and she usually links to the research in her slides). I read somewhere that Chinese scientists found it easier to get Chinese government funding for bold and ambitious research (and I imagine they’d also have other “advantages” given how the country is governed). So I wouldn’t be surprised if China would overtake the US in a whole bunch of technical sectors.

[ Parent | on HN ]

Re:U.S. widens trade blacklist to include some of China’s to... by BiasRegularizer on Wednesday October 09, 2019 @03:19AM (#21200850)

This is not completely true. China lead in terms of AI startup funding, but due to the huge shortage of talents, a good number of startups don't hold much water.

The top Chinese talents (e.g. Kaming He, Fei Fei Li etc) all chose to move to the US. AFAIK, there are no Chinese institutions that employ elite foreign AI researchers. The US universities and companies still lead with a large margin in terms of high impact publications in NIPS, ICML, CVPR etc.

[ Parent | on HN ]

Re:U.S. widens trade blacklist to include some of China’s to... by govg on Wednesday October 09, 2019 @04:31AM (#21201193)

You're wrong with the "large margin" terms. A lot of Chinese universities have stepped up in the recent past. You can take a look here : http://csrankings.org/#/fromyear/2017/toyear/2019/index?ai&v...

3 of the top 10 (all in the top 5 even) are Chinese in origin. This is a list of publications at the top venues in AI / ML for the past two years. If you extend it to past 5 years, you get a slightly better picture, but that doesn't change the 3 in top 5 comment either.

[ Parent | on HN ]

Re:U.S. widens trade blacklist to include some of China’s to... by BiasRegularizer on Wednesday October 09, 2019 @07:57AM (#21202303)

The keyword is "high impact". The link you provided is only ranked by publication count, it doesn't consider the impact factor of journal/conferences. If you look at stats from ICML this year[0], the only Chinese institution TsingHua is ranked #13 in the top #20 list.

Take a look at recent major advances in deep learning: information bottle neck, mask-rcnn, transformer, normalizing flow, all originated from US/Canadian/EU institutions. Chinese academia has very serious systematic issues that discourages innovation and encourages quantity over quality.

https://medium.com/@dcharrezt/icml-2019-stats-4ba18fbc6543

[ Parent | on HN ]

Re:U.S. widens trade blacklist to include some of China’s to... by govg on Wednesday October 09, 2019 @12:39PM (#21204882)

Sure, but at the same time there is a real bias towards established methods and research groups. It's hard to find traction as a newly established group doing original research - and "impact factors" are highly overrated since what ends up happening is the paper with most publicity and flash surrounding it is deemed more impactful. The fact that they went from almost zero representation at the international venues to such a high number should be indication that they are on the rise.

In addition, I think the sheer number of students and resources at their disposal will ensure China have a continuous pool of talent to draw from. The US has become increasingly hostile and competitive for ML researchers in terms of academic pressure, and returning to China has been incentivized for a while now.

[ Parent | on HN ]

Re:U.S. widens trade blacklist to include some of China’s to... by Linq123 on Wednesday October 09, 2019 @07:59PM (#21209504)

>Chinese academia has very serious systematic issues that discourages innovation and encourages quantity over quality.

Can you elaborate on that?

[ Parent | on HN ]