Google I / O developer
Feeling really nothing to say, is that Google released the latest smart phone Pixel 3a and Pixel 3a XL; released Android Q, and so on.
Microsoft's Build 2019 Developers Conference
1) cloud service Azure: Microsoft's most attention, the audience is very wide, serving a variety of Microsoft-based has its presence.
2) browser Edge: high fever, I basically every month to see related messages.
3) Windows Terminal: Terminal command-line tools, including PowerShell, cmd and WSL, WSL2 (linux kernel, personal feeling of little significance); wins developers might like it.
4) Microsoft 365: improving tools of Office, there is no breakthrough feature, tinkering with a lot of touches.
5) Microsoft Search: No personal sense.
On the AI industry
Mainly after reading "40 asked Kuang odd, as the CEO and India," the article, my feelings
(address: https://zhuanlan.zhihu.com/p/65560447)
量子位：你会认为算法是AI公司的核心壁垒或优势吗？
印奇：这关系到对算法不一样的定义。单一算法壁垒很低，但是能够系统化地产生大量算法，并且结合场景不断迭代、优化，最终还能打造算法相关的产品和方案，这个壁垒非常高。
量子位：2017年，你跟我们说过，2018年行业格局会落定，AI第一阶段的战斗将结束。
印奇：我认为已经结束了。当然从外部来看，会有个时间延迟。未来一两年，我们的核心竞争力，竞争对手的变化，会逐渐让大家看到这个结果。
量子位：怎么理解结束与否？
印奇：如果以跑步为例，这个阶段你可能不知道谁会跑到终点，但你会知道谁肯定跑不到终点。所谓第一个阶段，就像穿过一个筛子。从我们内部看，未来三到五年有能力推进到下一阶段竞争的公司，已经非常非常少了。
量子位：下一个阶段什么样？
印奇：如果把一个大行业和场景看成竞技场，接下来，会有很多新的、更有竞争力的玩家出现。你会发现，跟以前的小打小闹不一样，很多大家伙积累雄厚，牌怎么玩，变了。
我认为会变得更精彩。产业会慢慢整合，该倒闭的倒闭，该被并购的并购，该有希望的公司，就进入到下一个阶段。
Remember a long time before learning the word, probably means first technical, then the product, and finally to the market + management. And then direct it to say that for the AI industry, technology is not much of a problem, the product line (ie business) basically rolled out the next few years, we will start to grab the market, the loser will be merged, win It is not necessarily a victory, but based on some markets will have a considerable voice (the recent practice of this logic is in the Hundred Regiments and shared bicycle).
For the first 5,22,23 asked to sum up that is "core competencies / barriers is not a single algorithm or chip or other, but rather a system rooted in a particular industry, there is a sustainable iterative capability, an ecological system . "
5G
China Unicom officially released showed eight 5G industry scenarios, including "5G + unmanned" "5G + wisdom of the medical" "5G + wisdom environmental protection," "5G + Smart Energy," "5G + Industrial Internet," "5G + wisdom logistics" "5G + New Media", " 5G + port wisdom "and so on. Eight 5G industry has practical application scenario shows the scene outside the field of application of the exhibition hall and plaza. Guests from all walks of life together to visit and experience the innovative application of China Unicom in the field of 5G.
trade war
Personal feeling, the article misleading wall is very serious (Guo Jia from the point of view, this is not unreasonable), please polished eyes. Of those, "the United States to engage in daily Apple and other large companies", "United States and Europe each pinch" "US imperialism abandon its intention to die" like a fun look at the article on the line, do not take it seriously.
The so-called trade war, it is divided into four individual stages, first campaign, and then tear force of public opinion, and then sit down and negotiate, and finally landed execution. Currently in the first and second stages, (whether China and the US) pay attention to is to stimulate contradictory, and go wild speculations, anyway appeared very excited. But the third stage is the most important, this is the substance of things, at least to their own people seems a decent account of (the same as last year, hoping not to put forward a "victory" came out).
Baidu
Baidu first quarter net loss of over 300 million yuan, mainly to rising marketing costs, in the "Spring Festival red envelopes" invested 1.9 billion yuan.
With immediate effect resigned as senior vice president of Baidu to Dragon, the search firm executive officer (other March Ya-Qin Zhang declared retirement).
Robin Li said in the letter, as a leader, said, "We tried," to no avail, to ensure victory in a must win on the battlefield; every employee, at work to make every effort to ensure that every thing in place to perform .
Recently, the resignation of senior executives of many ah, to be a cool feeling.
ICLR 2019 summary (AI 2019 is the development trend of)
1) RNN sharp decline.
It is also expected, because although RNN applies to sequence data, but there is a great flaw: they can not parallel computing, and therefore can not take advantage of a larger study of drivers since 2012 - Operators force. RNN is not popular in the field of computer vision and reinforcement learning, once popular in the field of natural language processing, they are being replaced by attention-based architecture.
2) GAN momentum remains strong
) 3) reinforcement learning is still submit papers in the hottest topic of
research papers on ICLR 2019 General Assembly shows that reinforcement learning communities are never turned model approach and efficient model-based algorithm and meta-learning algorithm samples.
China, a bunch of big brother, big companies to speak, "pay attention to basic research, attention to innovation."
+ Guiding public opinion it is clear that the policy "include: Professor Yao said," can not seek immediate intensive research, the real foundation to do innovative research "; Tencent said farewell ism and so on.