With the release of Apple's new iPhone smartphones, iPhone 8, iPhone 8 Plus and iPhone X, almost everything we initially wanted to know is now almost clear, including retail prices, new naming, number of models, and many more. The latest features, etc., in fact, most of the information has been revealed before the press conference. Despite this, the new features of iPhone X, facial recognition technology, and AR reality enhancement technology are still very compelling.
However, there is a feature that Apple may not mention in the press conference or in the promotion. Many people may ignore its existence, that is, the powerful AI artificial intelligence feature.
Apple's three new models are equipped with a new mobile chip A11 bionics, Apple said the A11 bionic is the most powerful and intelligent chip on the iPhone. Regarding the "most powerful" point, we have learned from the benchmark data provided by yesterday's Geekbench 4 official that the A11 custom 6-core CPU performance is no match in the mobile field, leading all the Android flagship smartphones on the market, with The Snapdragon 835 leads more than 110% than a single core, and the multicore also has a lead of around 55%.
In fact, the performance of A11 is already a predictable thing, but where is the "smartest"? In fact, this has a very big relationship with the neural network engine integrated with A11. Although this part of Apple's press conference has been basically taken over and the publicity is low-key, it is undeniable that A11 is a SoC mobile chip for Apple's first neural network engine. Its neural network engine is a processing unit dedicated to running AI artificial intelligence. For this reason, Apple's naming of the A11 specifically highlights the AI ​​technical features through "bionics."
Privacy protection conflicting with the development of AI artificial intelligence
Before discussing the neural network engine, let's talk about the process of Apple AI artificial intelligence. As we all know, Apple is a company that is extremely privacy-conscious. Many previous examples are enough to prove this. Cook has even declared that Apple's confidentiality is higher than the US CIA.
Therefore, a large number of analyses believe that it is the impediment of privacy protection that has caused Apple to lose its initiative in AI artificial intelligence, which ultimately leads to the fall of the AI ​​artificial intelligence competition today. Apple not only lags behind other technology giants in making devices more user-friendly, but also has a low status in artificial intelligence cloud computing and has been marginalized for a long time. Why do you say that?
Deep learning is a sub-discipline of AI artificial intelligence. It usually needs to collect a large amount of data and aggregate it in the cloud for deep understanding and insight, but these practices are inconsistent with Apple's strict position on data protection. Apple uses end-to-end encryption technology in the FaceTIme and iMessage applications of iOS, and does not maintain or manage any personal data or information of users, which is consistent with Apple's strong view on consumer data protection.
Although Apple's position is commendable, it makes it difficult to develop its own AI, because Apple itself does not collect the big data needed to advance its AI artificial intelligence development, until now.
The new beginning of AI, not a new startIn today's mobile Internet era, data explosion, the emergence of big data makes AI artificial intelligence more and more useful. While data from many mobile devices can be analyzed in more depth in cloud computing, there are powerful AI computing solutions such as Google TPU and NVIDIA Volta. The problem is that data takes a while to reach the cloud, and Apple's consideration of privacy does not pass data to the cloud, so it's necessary to have a mobile device that provides slightly closer to the cloud's computing performance.
Apple explained at the press conference that there is an AI artificial intelligence called machine learning, which allows the device to learn by observation. Therefore, in a mobile SoC chip, there should be an AI processing unit dedicated to virtual neurons and deep learning. The neural network engine is a hardware developed for machine learning, which can not only perform high-speed operations required by neural networks. And it has outstanding energy efficiency.
In short, the neural network engine can take the task of CPU and GPU, greatly improve the computing efficiency of the chip, and complete more tasks faster with less energy consumption. In the A11 bionics, Apple's own neural network engine uses a dual-core design. The two cores are designed for AI artificial intelligence-specific machine learning algorithms. In real-time processing, the number of operations per second is up to 600 billion times.
Apple said that neural networks can run on CPU main processing units or GPU graphics processing units, but for this neural network type programming model, using custom chips for these applications, it will use a graphics engine when performing the same task. More energy efficient. The mystery of the neural network engine is its ability to handle matrix multiplication and floating-point processing, sharing specific tasks outside of the CPU and GPU, and achieving significant improvements in hardware performance.
In fact, Apple has long been aware of the trend of mobile AI processing units. A few days ago, Apple's senior vice president of hardware technology, Sweeney Sloki, said in an interview that when the iPhone 6 with the A8 chip was released three years ago, Apple was already developing the A11 biochip, but three years ago the mobile industry. There are not many topics about AI and machine learning. Schenny Sloki particularly stressed that "the built-in neural network engine is a bet that Apple has played in three years."
Neural network engine serves the present and the futureApple claims that the A11 with "Neural Network Engine Service" has become very smart, because it can transmit the neural network based on big data deep learning training to the mobile phone, and provide complete AI knowledge combined with the local neural network engine. ability. For local AI artificial intelligence processing, the current neural network engine has been able to perform many tasks, including smarter recognition of people, places and objects, providing powerful performance for innovative functions such as "face ID" and "moving expression". Wait.
The “face ID†for iPhone X applies AI artificial intelligence technology such as smart face recognition and machine learning. The original deep-sensing camera system uses the invisible infrared light to "illuminate" the person's face, and then the dot matrix projector projects more than 30,000 invisible spots on the face. To create a precise, unique depth map, the infrared camera detects subtle changes in the lattice reflection, reads the dot pattern, and captures its infrared image data. ?
The key is that the depth data of these very accurate infrared images and dot patterns captured will be sent to the neural network in the A11 biomimetic chip to create a mathematical model of the face, and then these can be accurately mapped. The data is sent to the security compartment to confirm that the data matches. Apple never transmits biometric data to the network, as does the data for the face function.
At the same time, the face ID function is based on the AI ​​artificial intelligence technology of the neural network engine. Even if the appearance of the face changes with time, it can be adjusted accordingly, even if it is makeup, makeup remover, glasses, hat or You can recognize it at the same time when you leave your beard. It is generally impossible to use the deceptive means of photo or mask to break the face ID, which is basically impossible under AI technology. ?
In addition to serving the face ID, AI technology also benefits from the camera system of the new iPhone. For example, the portrait mode self-timer now has the AI ​​technology blessing of the neural network engine, the depth of field blur effect is more prominent, and the extended "personal light effect" feature also utilizes the AI ​​complex algorithm to calculate the appearance characteristics. Affected by light, using data to create outstanding light effects. As for the “moving expressionâ€, AI technology allows the captured motion to analyze muscle movement more accurately, in order to reproduce the look.
Of course, A11's neural network engine is the core of Apple's AR augmented reality experience and Siri's personal assistant, and will be used in more areas in the future, and will be extended to more areas of the Apple ecosystem, including medical health-related applications, People driving car systems, Apple Watch, Apple TV and HomePod speakers, so that more blunt devices can interact in the same way as humans, and become really thinking.
In the case of privacy, Apple AI will continue to improve?
Apple is one of the earliest manufacturers in the mobile phone industry to enter the AI ​​artificial intelligence field. As early as 2011, it has integrated the Siri Assistant for the first time in the iPhone to provide corresponding speech recognition support. The appearance of the A11 biomimetic chip is another major move in the field of AI artificial intelligence. Its neural network engine marks a new beginning. Although Apple AI artificial intelligence is still lagging behind other competitors, using the new approach of neural network engine, Apple can accelerate its development without infringing on user privacy.
A11 Bionic chip integrated neural network engine, mainly for local large-scale data processing, with the chip's own powerful processing capabilities, greatly enhances cognitive ability, provides users with the most direct service locally, and the security of local processing of privacy data is beyond doubt. In the case of strong local AI processing power, it can cooperate with the development of cloud AI data for many years to further provide a complete and efficient experience for the device.
In any case, in the new model of mobile SoC integrated AI artificial intelligence processing unit, Apple has a greater advantage than other competitors, the focus is the soft and hard strength of the Apple industry benchmark and mature ecosystem. Many AI-based functions, such as face recognition, text analysis, and speech recognition, allow developers to more quickly apply it to the app to seamlessly switch between CPU and GPU to provide maximum performance and efficiency.
At the same time, thanks to the huge ecosystem and equipment technology, once an application is completed, it can be quickly expanded to each iOS device, expanding the scope of application, and further demonstrating Apple's unique product experience with AI. Apple said, "iPhone X unlocks the next decade of the iPhone", I believe that in addition to the full screen and face ID, AI neural network engine will be one of the key points.
GPS /Glonass/Beidou/Galileo Antenna
1. According to the polarization mode, GPS Antenna can be divided into vertical polarization and circular polarization. With current technology, vertical polarization is not as effective as circular polarization. Therefore, GPS antennas will be circularly polarized except in special cases. 2. GPS antenna is divided into internal antenna and external antenna in the way of placement. The location of the antenna is also important. At this time, the antenna is basically isolated from the interior of the whole machine. EMI is almost not affected. The satellite reception effect is very good. Now, with the trend toward miniaturization, GPS antennas are mostly built in. In this case, the antenna must be higher than all metal components, and the shell must be electroplated and well grounded, away from EMI interference sources, such as CPU, SDRAM, SD card, crystal oscillator, and DC/DC. The use of GPS in cars will become more common. And the car casing, especially the car explosion-proof film can seriously block the GPS signal. An external antenna with magnets, which can be attached to the roof, is essential for onboard GPS. GPS antenna structure
GPS Antenna,GPS Antenna for Car,GPS Antenna SMA,GPS Antenna Fakra,Inside GPS Antenna
Yetnorson Antenna Co., Ltd. , https://www.xhlantenna.com