[Guide] From November 1st to 5th, the 18th China International Industry Fair was held at the National Convention and Exhibition Center (Shanghai). Among them, the robot exhibition has always been one of the exhibition projects that attract the audience's attention most at the CIIF, and this year is no exception. Artificial intelligence has always been a pioneer in robots, so let us re-understand artificial intelligence.
Artificial Intelligence (ArTIficial Intelligence), the English abbreviation is AI. It is a new technological science that studies and develops theories, methods, technologies and application systems used to simulate, extend and expand human intelligence.
Artificial intelligence is a branch of computer science. It attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a similar way to human intelligence. Research in this field includes robotics, language recognition, image recognition, Natural language processing and expert systems, etc. Artificial intelligence is a simulation of the information process of human consciousness and thinking.
Artificial intelligence is not human intelligence, but it can think like humans and may exceed human intelligence. Artificial intelligence has now become the world's three cutting-edge sciences, alongside space technology and energy technology.
This company is awesome! On October 31, according to foreign media (cnbc) reports, Graphcore, a startup that produces AI chips, raised $30 million in funding. These funds will help it compete with the traditional giants Intel and Nvidia in the future chip field.
The British company plans to ship on a large scale next year, and its chips will be used in driverless cars and cloud computing. These fields mostly use machine learning technology, and there are a large number of machine learning calculation requirements.
Graphcore claims that regardless of the speed of AI learning or energy consumption, its chips are 100 times ahead of similar products in the market.
Nowadays, graphics processing units (GPUs) are often used to run machine learning programs. Graphcore CEO Nigel Toon said that this is just a stopgap measure for the industry and is not suitable for future AI computing needs.
Toon said in a telephone interview on Monday: "GPUs are designed to run complete programs, and machine learning is completely different. The latter is the process of continuously training programs to use data and requires completely different types of processors." Graphcore A new Intelligent Processing Unit (IPU) system is currently being built and is scheduled to be launched in 2017.
The investors in this financing are led by Robert Bosch Venture Capital, and also include Samsung, venture capital firm Amadeus Capital Partners, C4 Ventures, Draper Esprit, FoundaTIon Capital, and Pitango Venture Capital.
In the past two years, Graphcore has been committed to the research of AI chip technology. Toon claims that its technology can help reduce costs and increase efficiency. Toon gave a case of a social media company. The user activity of social networks is usually the most prominent at a certain time. Graphcore's IPU system will wait for the opportunity to use idle processor resources for AI training when user activity is low, and then perform AI training in the first New content will be launched in a timely manner within two days.
According to marketing intelligence company TracTIca, the deep learning market has huge potential. Its construction expenditure was US$436 million in 2015 and will grow to US$41.5 billion by 2024. The market is currently dominated by Intel and Nvidia. The two companies respectively introduced new processors designed for AI programs earlier this year.
In the initial stage of the business, Graphcore's main goal is to sell products to companies interested in training AI systems. However, Graphcore also said that it is already discussing the next round of product applications with strategic investors.
"We have Bosch and Samsung as strategic investors. Bosch is interested in autonomous vehicles and new types of transportation, while Samsung is interested in next-generation network equipment." Toon said. "We are discussing the development of these projects with our partners."
Attached are 5 companies dedicated to AI chip research:
1. KnuEdge
KnuEdge is not actually a start-up company. It was founded by the former head of NASA and has been operating in a stealth mode for 10 years. KnuEdge recently stepped out of the invisible model and let the world know that they received $100 million in investment from an anonymous investor to develop a new "neuron chip."
KUNPATH provides chip technology based on LambaFabric. LambaFabric will perform neural network calculations through a completely different architecture from GPUs, CPUs and FPGAs currently on the market. LambdaFabric is essentially designed to scale up to 512,000 devices in a highly demanding computing environment, with a rack-to-rack delay time of only 400 nanoseconds, and a low-power 256-core processor. KNUPATH technology is based on biological principles and will redefine chip-level/system-level computing in the data center and consumer equipment market.
Compared with other similar chips, this chip technology should provide 2x to 6x performance advantages, and the company has already gained revenue by selling their prototype systems. Under the "KnuEdge umbrella structure", KnuEdge is composed of 3 separate companies. KnuPath provides their chips, KnuVerse provides proven military-grade speech recognition and verification technology, and Knurld.io is a program that allows developers to Public cloud API service (Public cloud API service) that integrates voice verification into their patented product. KnuEdge claims that it is now possible to verify computers, networks, mobile applications and IoT devices by just saying a few words into the microphone. What a great thing it would be to never have to remember your password in the future?
2. Nervana
Founded in 2014, Nervana Systems, a San Diego-based startup, has received US$24.4 million in funding from 20 different investment institutions, one of which is the highly respected Draper Fisher Jurvetson (DFJ) . The Nervana Engine (coming out in 2017) is an ASIC chip customized and optimized for deep learning. The realization of this scheme benefits from a new memory technology called High Bandwidth Memory, which has both high capacity and high speed, providing 32GB of on-chip storage and 8TB of memory access per second. The company currently provides an artificial intelligence service "in the cloud", which they claim is the fastest in the world and is currently used by financial service institutions, healthcare providers and government agencies. Their new chip will guarantee The Nervana cloud platform will maintain the fastest speed in the next few years.
3. Horizon Robot
Horizon RoboTIcs (Horizon Robotics), a start-up founded by the Chinese in 2015, has received an undisclosed amount of seed funds from investors including Sequoia and legendary venture capitalist Yuri Milner. On July 1, 2016, Horizon Robotics received a new round of financing. This investment will be used to increase R&D investment in the field of autonomous driving and smart home, speed up product development and implementation; promote the research and development of artificial intelligence chips and systems . They are starting to build a one-stop artificial intelligence solution that defines "all things intelligence" to make life more convenient, more interesting, and safer.
Horizon is committed to building an artificial intelligence "brain" platform based on deep neural networks-including software and chips, which can achieve low-power, localized solutions to environmental perception, human-computer interaction, decision-making control and other issues. Among them, in terms of software, Horizon has made a neural network-based OS, and has developed the "Hugo" platform for autonomous driving and the "Andersen" platform for smart homes, and they have begun to land gradually. In terms of hardware, in the future, Horizon Robotics will design a chip for this platform-NPU (Neural Processing Unit) to support its own OS. By then, the performance will increase by 2-3 orders of magnitude (100-1000 times).
4. Vimicro
Speaking of domestic artificial intelligence chips, I have to mention that on June 20 this year, China Star Microelectronics was the first to launch China’s first embedded neural network processor (NPU) chip. This is the world’s first embedded video with deep learning artificial intelligence. Collect the compression coding system-level chip, and named it "Xingguang Intelligent No. 1". This deep learning-based chip is used in face recognition and can achieve an accuracy rate of up to 98%, which exceeds the recognition rate of the human eye. The chip was mass-produced on March 6 this year, and the current shipment volume is more than 100,000 units.
The NPU adopts a "data-driven" parallel computing architecture. A single NPU (28nm) consumes only 400mW, which greatly improves the ratio of computing power to power consumption. It can be widely used in high-definition video surveillance, intelligent driving assistance, and Embedded machine vision fields such as humans and robots.
5. krtkl
Krtkl, founded in 2015, is dedicated to creating "a tiny wireless computer used to create something completely different". Technical people will be obsessed with Snickerdoodle, a dual-core ARM processor, FPGA, WIFI, Bluetooth, starting at $65, "the smallest, most difficult and most affordable platform to empower robots, drones, and computer vision." This product actually received more than 160,000 US dollars through crowdfunding. The latest information is that they have received the primary version of Snickerdoodle and will ship soon
。
