With the coming of 2018, the technology of artificial intelligence will be further improved, but for the five major problems of artificial intelligence, experts will be racking their brains. If the existing AI technology is widely adopted, it will bring great changes to society.
In terms of a lot of hype about "killer robots", there have been some notable advances in artificial intelligence in 2017. Alpha dogs, cold-swing masters and other chess robots can make top players fall into despair. In the real world, machine learning is being used to improve agriculture and expand the coverage of health care.
But have you talked to Siri or Alexa recently? If there is, then you will know that, apart from these hype, and the smug billionaires, there are still many things that artificial intelligence can't do or understand. Here are five tough questions that experts will rack their brains for next year. Please see the small series for you to come together:
Machines do better than ever before in dealing with text and language. Facebook can read image descriptions for the visually impaired. Google has done a great job of giving short suggestions when replying to emails. However, the software still does not really understand the meaning of our words, or the ideas we want to share with them. Portland State University professor Melanie Mitchell said: "Humans can combine the concepts we have learned in different ways and apply them in new situations. AI and machine learning systems cannot. ."
Mitchell describes the problems facing today's software as what the mathematician Gian Carlo-Rota calls "meaning barriers." Some leading AI research teams are trying to figure out how to overcome it.
Part of this work is to provide the machine with a cognitive basis for common sense and the physical world – they set the way for us to think. For example, Facebook researchers are teaching software to understand reality by watching videos. Others are simulating what we can do with knowledge about the world. Google has been trying to build software that understands metaphors. Mitchell experimented with a system that uses analogies and concept storage to explain what is happening in the photo.
The "reality gap" that hinders the machine revolutionThe robot hardware has developed quite well. For $500, you can buy a palm-sized drone with an HD camera. The robot carrying the box and the robot walking on both legs have also been improved. So why are we not surrounded by busy mechanical assistants? Because today's robots lack a brain that matches their advanced muscles.
Letting the robot do anything requires specific programming for a specific task. They can learn through repeated trials (and errors), such as grabbing objects. But this process is relatively slow. A promising shortcut is to let the robot train in a virtual, simulated world and then download those hard-won knowledge to the physical robot. However, this approach is plagued by real-world disparities. Specifically, the skills that robots learn during the simulation process are not always effective when moving to machines in the physical world.
This reality gap is shrinking. In October, in an experiment where virtual and real robotic arms picked up a variety of items—these tasks included tape dispensers, toys, combs, and so on—Google reported promising results.
For autonomous vehicle practitioners, it is important to make further progress. In the machine driving competition, many companies are deploying virtual vehicles on virtual streets, and they hope to reduce the time and money spent testing in actual traffic and road conditions. Chris Urmson, CEO of autonomous driving startup Aurora, said that making virtual testing more suitable for real vehicles is one of the team's priorities. Urmson, who led the independent car project of Google's parent company Alphabet, said: "We can use this technology to accelerate learning next year or later."
Jasminer Machine:Jasminer X4,Jasminer X4-Q,Jasminer X4-1U,Jasminer X4-C 1U,Jasminer X4 BRICK
As a leading brand in the high-throughput server industry, JASMINER has always placed scientific and technological research and product innovation in the highest strategic position and has no upper limit on R&D investment. It is the dedication to technology research and development that enables JASMINER to take the lead in the field of energy saving technology.
At present, JASMINER is one of the world's few cutting-edge computing server brands that apply the core chip technology of "integrated storage and computing". With the chip stacking technology, JASMINER X4, the industry's leading high-throughput computing chip of integrated storage and computing, has overcome Moore's law and led the world. Thus, the energy saving, stability and reliability of the computing power server are further improved.
Based on energy saving, consumption reduction and comprehensive cost optimization, JASMINER X4 series server products have gained strong competitiveness in green environment protection, cost reduction and efficiency improvement for global customers. X4 products based on high-throughput chip architecture, with strong computing performance and excellent energy efficiency, fully meet the needs of green computing, for the development of the new generation of information technology to provide a powerful new data infrastructure guarantee.
Jasminer Machine,X4 1U Etc Miner,jasminer miner,X4 1U 520Mh Miner,jasminer x4
Shenzhen YLHM Technology Co., Ltd. , https://www.sggminer.com