News on China's scientific and technological development.

Wuhun

New Member
Registered Member
If I understand what they are saying about this quantum computer stuff correctly, then they are there or right there on the line of completing a full operational and programmable quantum computer.
No. All the frontier quantum computing teams are at around ~100 "physical qubits" with only Xanadu >200 qumodes [
Please, Log in or Register to view URLs content!
] The key difference

Qumodes = infinite dimensional continuous variable system; though currently they can only stack up to 3-dimension.
Qubits = 2-dimension system; so X number of qumodes >> X number of qubits

However, Xanadu only demonstrated their capabilities in "Boson Sampling" which USTC has already evinced [
Please, Log in or Register to view URLs content!
] using 76 photon qubit. So far no one has been able to demonstrate any real-world application using quantum computer. That Google claim of so-called "quantum supremacy" was fraudulent as News on China's scientific and technological development.

Now what the Huazhong team has demonstrated is a new revolutionary technique to build quantum logic gates using "linear-optical gate" instead of "two-photon logic gates" with entangling gate fidelity of 99.7%.

Alibaba currently has the highest fidelity in superconducting circuits -
1659945038076.png
Now to build practical quantum computer one must have at least 10000 physical qubit for 1 logic qubit (for general purpose quantum computer ~ 1 million physical qubit) and this where error-correction comes into play. So, 10K physical qubit are far away as to built them error-correction is needed which require quantum algorithms with deep math underlying quantum mechanics called surface code. The 3 leading teams are

* USTC -
Please, Log in or Register to view URLs content!

* ETH Zurich -
Please, Log in or Register to view URLs content!

* Google -
Please, Log in or Register to view URLs content!


That super conducting quantum computer people are working on (the frozen apparatus)
Well no matter what approach one takes whether-
1. Photon
2. Superconducting circuits, transmon or fluxonium
3. Ions
4. Cold atom
5. Topological majorana fermion
6. Electron Spin
7. Nuclear Spin

cryogenic dilution refrigeration is used in atleast one of the process. For photonic quantum computer, the photon detector needs cryogenic apparatus.
 

Wuhun

New Member
Registered Member
Not quite quantum computing or supercomputing achievements, definitely a breakthrough in high-level math and coding.
I've alluded that,
Quantum mathematical theory [aka high-level math] and computational complexity research [aka coding] in China are now at the frontiers

Now, America for months have had exa-scale supercomputer, and the primary to do things you do with "peak supercomputer" is testing the existing limits of molecular dynamics, fluid mechanics, and/or "quantum simulation", and then publish the results, and win the respect or accolades like ACM Bell Prize. The notion that researchers didn't wanted to test the limits of their quantum computing tech may be disingenuous. The logical reasons are either -

1. They didn't possess the required quantum mathematical theory and computational complexity "frontier" capability, which can't be true because America is indeed at the frontier, therefore

2. America's super-computing performance may be behind that of China's. So the allusion that
Performance of China's Exa-flops supercomputer > America's
 

Overbom

Brigadier
Registered Member
Womp womp
Please, Log in or Register to view URLs content!

Chinese scientists develop flying unmanned submarine​

  • Propeller-driven drone has civilian and military applications, including inspecting underwater mines
  • This type of vessel is one of the cheapest and most effective methods for taking down the defence system of an air carrier fleet, according to researcher
Driven by four propellers, including a pair at the front that tilt, the drone can approach an underwater target slowly and linger in one area for a long time, they said.
Two large wings that fold over its back can extend when the drone reaches the surface of the water, allowing it to fly at a speed of 120km/h (74.6mph), about twice as fast as an ordinary drone powered by rotor blades.
The drone “consumes little energy when cruising in fixed-wing mode, so it can perform some fast, long-range missions in the air”, said professor Ang Haisong, lead project scientist with the Nanjing University of Aeronautics and Astronautics in a paper published in the peer-reviewed Unmanned Systems Technology in June.
A researcher studying similar technology at Northwestern Polytechnical University in Xian, Shaanxi province, said China was developing numerous types of “transmedia vessels” that could travel in both air and water.
“They are mainly for military applications. Some can fly at supersonic speed,” said the researcher, who was not involved in Ang’s study but requested not to be named because of the sensitivity of the technology.
Ji Wanfeng, professor with the Naval Aviation University in Yantai, Shandong province, said this type of transmedia vessel was one of the cheapest, most effective methods to cripple the defence system of an aircraft carrier fleet.
The multilayered defence system of a modern warship can shoot down about half of the approaching planes, missiles or traditional drones, according to an estimate by Ji and his colleagues.
But a transmedia vessel can go underwater when detected by radar and resurface to dodge sonar. Just a small number of these drones could confuse or even overwhelm a warship’s computer, Ji said.
A transmedia vessel launched from a distance of 100km (62 miles) had a survival rate of nearly 100 per cent if it could fly at a speed of more than 150km/h, Ji’s team said in a paper published in the Chinese journal Electronics Optics & Control last month.
This type of drone “can achieve efficient strikes against the enemy’s key targets. It will surely become a powerful supplement to the Chinese navy’s existing equipment combat methods and tactics,” Ji’s paper said.
The flying submarine developed by Ang’s team has an airbag in its belly that can be filled with water to adjust buoyancy so it can stay at a given depth without making noise with its propellers. When cruising in water with its wings folded, its streamlined body resembles that of a typical submarine, allowing for less drag and higher mobility.
The biggest challenge occurred during take-off. According to Ang’s team, if the vessel rises directly out of the water, the take-off process is unstable because of surface waves and the drone’s simultaneous interaction with air and water.
To fix this, they developed a complex control mechanism so that the submarine can glide on the waves before lifting.
 

luosifen

Senior Member
Registered Member
Please, Log in or Register to view URLs content!

2022-08-08 16:26:25Ecns.cn Editor : Zhang Dongfang
Please, Log in or Register to view URLs content!


Photo shows the exhibition hall of the 2022  China International Fair for Trade in Services at the Shougang Park in Beijing, August 1, 2022. (Photo/China News Service)

Photo shows the exhibition hall of the 2022 China International Fair for Trade in Services at the Shougang Park in Beijing, August 1, 2022. (Photo/China News Service)

(ECNS) -- Futuristic products including the first 5G robot dog for building inspection in China, on-site recorder, robot guide, and intelligent safety helmet will be unveiled at the upcoming 2022 China International Fair for Trade in Services (CIFTS) in Beijing.
These exhibits have combined cloud computing, Internet of Things, mobile internet, artificial intelligence and other advanced technologies.
With strong adaptability to various terrains, the 5G robot dog can replace humans to enter dangerous situations and realize remote automatic inspection task issuing and starting control.
It supports remote communication and can solve on-site problems in real time after discovering problems through intelligent scene identification.
The CIFTIS has set up a good platform for enterprises that want to go abroad and provided them with opportunities of cross-industry exchanges, a head of Glodon Company said.
Youdao’s intelligent products for study such as AI learning machine and smart dictionary pen will also be on display at the fair.
The CIFTIS consists of a comprehensive exhibition and special exhibition, covering an area of 152 thousand square meters. It will be held from Aug. 31 to Sep. 5 at Beijing Shougang Park.
 

Wuhun

New Member
Registered Member
Recently CVPR, one of the two most prestigious conference in "machine vision" was held at New Orleans. China had almost half of the accepted papers. View attachment 93797

Last time when the conference was held in person in 2019 at Seoul
View attachment 93798

Sensetime is probably the world's greatest computer vision company. Megvii is also a serious company.



CVPR2022 institutions.jpg

22. Huawei = 33

The entire list is here
Please, Log in or Register to view URLs content!
 

Wuhun

New Member
Registered Member
It appears that Chinese AI researchers and institutions focus heavily on CV relatively speaking. In comparison, in the US AI researchers weigh more heavily to NLP. Not sure if it has a lot to do with the wide acceptance and popularity of face recognition.

Your argument is full of fallacies as

1. Computer vision =/= face recognition.

2. To build General AI aka Super-intelligence, the importance of computer vision is same as NLP.

3. Biggest application of computer vision
  • Automation - autonomous vehicles, factory/industrial automation, agriculture​
  • Medical AI​
4. China had ~25% papers in last ACL which is the most prestigious NLP conference

5. China currently has the largest large-language model aka foundation model aka Superintelligence algorithm [BAAI] which is Academia with no American equivalent. By companies
  • America - Google, Microsoft+OpenAI, Microsoft+NVIDIA, Meta/Facebook
  • China - Huawei, Baidu, Alibaba, Inspur
6. ~50% computer vision & NLP research in America are done by the Chinese
 

weig2000

Captain
Your argument is full of fallacies as

1. Computer vision =/= face recognition.

2. To build General AI aka Super-intelligence, the importance of computer vision is same as NLP.

3. Biggest application of computer vision
  • Automation - autonomous vehicles, factory/industrial automation, agriculture​
  • Medical AI​
4. China had ~25% papers in last ACL which is the most prestigious NLP conference

5. China currently has the largest large-language model aka foundation model aka Superintelligence algorithm [BAAI] which is Academia with no American equivalent. By companies
  • America - Google, Microsoft+OpenAI, Microsoft+NVIDIA, Meta/Facebook
  • China - Huawei, Baidu, Alibaba, Inspur
6. ~50% computer vision & NLP research in America are done by the Chinese

My short comment assuming a lot of contextual knowledge appears to be easily misunderstood. Face recognition is one of the most common applications of CV in China and might have played a role in the popularization of CV. It's of course not the only application of CV. I was also talking in relative sense, which is clear in my comment.

I'm very familiar with everything you mentioned above and more.
 

Wuhun

New Member
Registered Member
My short comment assuming a lot of contextual knowledge appears to be easily misunderstood. Face recognition is one of the most common applications of CV in China and might have played a role in the popularization of CV. It's of course not the only application of CV. I was also talking in relative sense, which is clear in my comment.

It appears that Chinese AI researchers and institutions focus heavily on CV relatively speaking.

Chinese AI institutions
China had ~25% papers in last ACL which is the most prestigious NLP conference

Chinese AI researchers
~50% NLP research in America are done by the Chinese

US AI researchers weigh more heavily to NLP

China currently has the largest large-language model aka foundation model aka Superintelligence algorithm [BAAI] which is Academia with no American equivalent. By companies
  • America - Google, Microsoft+OpenAI, Microsoft+NVIDIA, Meta/Facebook
  • China - Huawei, Baidu, Alibaba, Inspur

Not sure if it has a lot to do with the wide acceptance and popularity of face recognition.

2. To build General AI aka Super-intelligence, the importance of computer vision is same as NLP.

3. Biggest application of computer vision
  • Automation - autonomous vehicles, factory/industrial automation, agriculture​
  • Medical AI​


I'm very familiar with everything you mentioned above and more.
Since you're "familiar" then put on more effort
My short comment assuming a lot of contextual knowledge appears to be easily misunderstood
 
Top