This article is produced by NetEase Smart Studio (public number smartman 163). Focus on AI and read the next big era!
Editor's note: On October 24th, the artificial intelligence startup Shen Jian Technology, which was established for just over a year, held a new product launch conference in Beijing. At the press conference, Shenzhen Shen Technology announced six deep learning products and a 2018 chip R&D plan.
At the press conference, SCI Technology also announced that the company has completed approximately A$40 million A+ round financing. The financing was led by Ant Financial and Samsung Ventures, followed by China Merchants Venture Capital and China Venture Capital. Yao Song disclosed that this round of financing will continue to be used in security, data center and subsequent chip product development. In just over a year, Shen Jian Technology has completed three rounds of financing.
After the meeting, NetEase Intelligence was invited to interview the three founders of Shenzhen Sci-Tech Co., Ltd. (CEO Yao Song, CTO, and co-founder Wang Yu) and asked them about their opinions on artificial intelligence chips, future application fields, and Why did you choose to enter the market in the security field?
Text
Released six new products. Why did you go deep into the field of security?
Shen Jian Technology's six products released this time are Video Structured Solution (DP-2100-O16), Face Analysis Solution (DP-2100-F16), Face Detection and Identification Module (DP-1200- F1), a deep understanding of Aristotle architecture platform, deep depth learning and development SDK (DNNDK) and deep voice recognition acceleration program (DP-S64). The first five are video surveillance applications and related solutions.
It is worth mentioning that DNNDK (Deep Neural Network Development Kit) is the first domestic original deep learning development SDK developed by Shenzhen Sci-Tech for the AI ​​heterogeneous computing platform DPU (Deep-Learning Processor Unit). . At the press conference, the CTO demonstrated the workflow and principles of the face recognition program for the audience, and also demonstrated the deep-learning SDK model.
Shen Jian Technology CEO Yao Chung said that DNNDK can greatly reduce the DPU platform depth learning application development threshold and deployment difficulty, and greatly shorten the algorithm deployment to the hardware cycle, speed up AI product development process. “DNNDK is the first deep learning development kit in China and can be used to benchmark the Tencent RT.†Yao Song believes that as a start-up company, Shenzhen Sen Technology must focus on the scenarios and solutions, and combine the vertical industry to do business development. It may be difficult to face up with Nvidia as a purely generic machine learning chip.
As for why use AI technology to enter the field of security? Yao Song's explanation stated that two factors were mainly considered. The first was the total market volume. Yao Song said, “We did drones last year. It is a very good early validation scenario. It is not very sensitive to performance, power consumption, and value. But the total market is too small. This year we decided Abandoned.†Another factor is to consider the time when the market begins to mature. “We have not taken auto-driving as our core to the company. There are still some robots that are not considered core, because we think that the market has The income will be a bit slower," Yao Song believes that the security industry is now able to see a large amount of income in the industry.
"In addition, there is a secondary factor that is the competitive landscape of the industry. Assuming that 80% of the industry is a giant monopoly, it may be worthwhile for everyone to do this industry. Automated driving and security are very separate markets. There will be a great chance to do it," Yao Song said.
According to Yao Song in an interview, currently about 70% of SZM’s revenue comes from security and the rest comes from the data center.
To be a platform-level chip, in the vertical industry
At the press conference, Shen Jian also announced the self-developed deep-learning SoC “Tingtaoâ€, which will be installed in Shenzhen Jian’s solution in the first half of 2018.
Looking at the market, many companies currently roll out many chips based on artificial intelligence technology. Many of them are followed by big vendors like Google, Microsoft, and Facebook. All companies are making their own chips.
Not only that, the original chip company also began to seek change, Nvidia originally made GPU, and later it created the Internet ecology, and now it provides complete machine solutions, as well as a complete solution with automated driving algorithms and medical imaging solutions. In Yao Song's opinion, the entire chip industry has a vertical level of communication, AI's impact on the industry is very great.
From a single point of view, AI technology spirals upward. "When you emerge from a certain point of view, such as the breakthrough of AI algorithm, both the chip and the whole industry chain technology will increase." Specifically for the chip industry, With the entry of AI, it will provide a lot of new requirements and put forward many improvements from the chip technology, including chip manufacturing, chip packaging testing, will have an impact on the entire industry.
Singh said at the same time that it is not possible to make chips for chips, but to increase performance and reduce costs. "If we don't find a scenario that is very large, it is actually a waste of money to make chips." Singh said that the chip has a very high one-off investment. This investment requires a huge amount of shipments to dilute. In order to be able to have a large amount, it is necessary to be a platform-level chip. "It will have a The entire software development environment can support different resolutions, different applications, different networks, different segmentation, and detection of certain tasks."
However, the development of AI chips is still at an early stage and there is currently no industry standard. Shen Jian Technology CTO single nickname, we need an industry standard to assess the AI ​​chip. "The first one is whether it's a real AI chip because many of the so-called 'AI chips' are just custom acceleration for a particular industry-specific algorithm. At this point, we are more inclined to us if we can find some Typical test cases, including performance power consumption, testing of typical networks, etc., can filter out many so-called “AI chips.†Dan said, “We hope that the sooner this standard emerges, the better. We hope that this standard will be established. More rigorous, the more test cases are inside, is beneficial to us. â€
Expand multiple application scenarios, mobile phone AI chip war or will start
With this round of financing, Shen Jian Technology has had more endorsements from major manufacturers. Thanks to the strategic resources injected by Ant Financial Services, Yao Song said that he will further develop more application scenarios including finance, and Shen Jian will also cooperate with the investor Samsung in various aspects including storage.
As the largest chip-using device of the order of magnitude, mobile phones appear to be ushering in more and more AI applications. Yao Song believes that the application of deep learning in mobile phones will certainly be particularly special. In this regard, Apple generally leads the trend in this field.
As for how to layout, Yao Song expressed his own view: "I think the computing power provided by mobile phones is limited. It is really necessary to do a very fine understanding of the scene or analysis, etc., or it needs more computing power. Therefore, many manufacturers often send some things to the cloud and do more detailed analysis. In terms of AI applications, cloud and endpoints must be coordinated and will not be completely confined to mobile phones.
As for the depth of Kam-Hon Tao chip is not possible on the mobile phone side, Yao Song said there is hope. He said that the initial launch of the HeTuo chip power consumption is about 1.1W, a few T digital chips, this chip can be tailored to suit lower power consumption scenarios. For example, the power consumption of the Internet of Things chips is usually below 100 milliwatts. Inside the mobile phone is 100 to 500 milliwatts, and in security we want it to be 1 watt.
As for the platform class chip is not meant to sacrifice performance? Shan Jian said that Shenzhen Shen Technology will use its own powerful algorithm to compress the model. The chip is designed to be completely customized for deep learning. Although it is platform-level, it guarantees customization in the field of customization. The efficiency. Shan Yan pointed out that the efficiency of the chips designed by the Shen Jian team is much higher than that of Apple's AI chips specifically designed for face recognition, as well as the efficiency of the Huawei Unicorn 970. As for the future, Shenjian will not create AI chips for Samsung handsets, and it has not yet been confirmed by Yao Song.
Pay attention to NetEase smart public number (smartman163), get the latest report of artificial intelligence industry.
ZGAR Vape Pods 5.0
ZGAR electronic cigarette uses high-tech R&D, food grade disposable pod device and high-quality raw material. All package designs are Original IP. Our designer team is from Hong Kong. We have very high requirements for product quality, flavors taste and packaging design. The E-liquid is imported, materials are food grade, and assembly plant is medical-grade dust-free workshops.
From production to packaging, the whole system of tracking, efficient and orderly process, achieving daily efficient output. WEIKA pays attention to the details of each process control. The first class dust-free production workshop has passed the GMP food and drug production standard certification, ensuring quality and safety. We choose the products with a traceability system, which can not only effectively track and trace all kinds of data, but also ensure good product quality.
We offer best price, high quality Pods, Pods Touch Screen, Empty Pod System, Pod Vape, Disposable Pod device, E-cigar, Vape Pods to all over the world.
Much Better Vaping Experience!
Pods, Vape Pods, Empty Pod System Vape,Disposable Pod Vape Systems
ZGAR INTERNATIONAL(HK)CO., LIMITED , https://www.szvape-pen.com