The outside world has always thought that Apple is relatively low-key in the field of artificial intelligence, and its technology is also behind Google, Microsoft and other companies.
In order to allow the public to understand Apple, Lei Feng network (search "Lei Feng network" public concern) yesterday released "Apple's artificial intelligence the most complete decryption, iBrain already everywhere" article on Apple artificial intelligence in-depth analysis.
The book has revealed that Apple has written two pages of machine learning applications, some of which are already in use, and others are still under discussion. Behind these applications is that Apple has quietly acquired 15 artificial intelligence companies over the past 6 years, including innovative technologies such as speech recognition, natural language processing, image recognition, face recognition, motion capture, and machine learning. the company.
An interesting phenomenon is that after each acquisition of an AI company, Apple will not publicly announce the acquisition purpose and development plan, and will immediately close the products and services of the acquired company.
 Speech, semantic recognition 
Siri Voice Assistant
Siri is undoubtedly Apple's most important voice product. Buying it at a price of US$200 million in 2010 was undoubtedly a very cost-effective transaction. Afterwards, Apple’s acquisitions in the direction of voice recognition and NLP were based on Siri.
Novauris Speech Recognition
In April 2014, Apple acquired Novauris, an automated voice recognition technology company. Novauris was founded by Nuance, the founder of Dragon Systems, and Nuance was the basic voice technology provider for Siri. Their core product is the server-based NovaSystem, a scalable speech recognition system that features simultaneous processing of multiple voice access requests. Although Apple declined to disclose how to use the Novauris team, from the perspective of the relationship between Siri and Nuance, Apple's acquisition is intended to get rid of Nuance's dependence and use the Novauris team to develop its own voice technology.
VocalIQ human-computer dialogue
All along, Siri can only identify the most basic instructions and can't make high-matching answers, which makes it an entertainment product for entertainment. In October 2015, Apple resolved this issue by acquiring VocalIQ. VocalIQ can use deep learning to understand the language environment and make human-computer conversations more natural. After embedding the artificial intelligence technology of VocalIQ in Siri, the developer uses this platform to store and learn user's communication information, to accurately identify the user's instructions and provide more intelligent dialogue.
 Machine learning 
Apple once mentioned that they use AI technology to test some micro-functions: such as identifying strange calls; detecting the user's movement status; listing the most likely to use applications after the mobile phone is unlocked; automatically displaying nearby marked hotels; scheduling; interest news recommendation and many more. These seemingly simple functions were all put into the machine learning system. For this reason, Apple used its machine learning company, which was interested in recommendation and habit prediction, to increase its own data mining capabilities.
Turi machine learning platform
In August 2016, Apple spent $200 million to acquire Turi, a machine learning company. Currently Turi has launched GraphLabCreate, Turi machine learning platform, TuriDistributed, and TuriPredictive Services, which are mainly used to develop solutions such as recommendation engines, sentiment analysis, and fraud detection.
Cue data mining
In October 2013, Apple acquired Cue, a search engine called “cloud data,†for US$40 million. Cue can be accessed from user emails, contacts, Facebook, Twitter, LinkedIn, Reddit, Dropbox, Evernote, and Tumblr. Collect data, organize all the data, and find out the information the user needs through machine learning algorithms, such as providing calendar reminders, notifications and appointments, meeting restaurants, etc. It is reported that Cue will be integrated into Siri.
Matcha video recommendation
In August 2013, Apple acquired Matcha for approximately US$10-15 million (the two parties did not publish a clear transaction price), and its products can be captured from streaming media sites such as Netflix, iTunes, Hulu, and Amazon Prime. The right information to the user. Apple's acquisition of Matcha is mainly to obtain video recommendation algorithms. Prior to going offline, Matcha’s iOS app was among the top 15 in the App Store entertainment category’s best-selling software, and its user growth rate was very fast.
Semetric music data mining
In January 2015, Apple acquired Semetric Corporation for US$50 million. The company launched Musicmetric Service in 2008. Its main business is to provide data analysis services related to music download and streaming, and use machine learning to help customers analyze music in social media. Which singers and songs are loved by the fans, and thus give relevant music recommendations. After the acquisition was completed, Apple integrated it into iTunes music. In addition to music, Semetric's data analysis services for games, television, movies, and books also have accumulated, which will help enhance Apple's data mining of its various digital products.
Spotsetter map social personalized recommendation
Spotsetter is a social search engine based on Google Maps. It was acquired by Apple in June 2014. It analyzes users' social circle data, provides users with personalized location recommendations (such as travel locations and restaurants), and is committed to solving so-called "where to go" problem. The purpose of Apple's acquisition is nothing more than the prospect of its recommendation algorithm and Apple Maps integration.
WiFiSlam Machine Learning and Pattern Recognition
In March 2013, Apple acquired WiFi Slam, an indoor navigation service provider, for US$20 million. The company’s strength lies in machine learning and pattern recognition technology. It can correlate the data collected by various sensors on the device, and is positioned with the WiFi triangle. The data is combined to draw an accurate indoor map and Apple applies its algorithm to the map.
Topsy social data mining
In 2013, Apple acquired Topsy for $200 million. The company's products can help users extract key information from Twitter and other social media data, including keyword trackers and determine how users in social media respond to a topic.
 Computer vision 
Apple officials mentioned that their face recognition and video detection technologies have begun to be used in cameras and other products. At the same time, sources said that Apple has acquired several companies specializing in computer vision for the development of VR/AR products.
Perceptio image recognition
In October 2015, Apple acquired image recognition company Perceptio. The company's product is the development of smart phone image classification system for artificial intelligence. The biggest advantage of this system is that it does not require a large amount of external data for classification. This is in line with Apple’s privacy protection strategy, which minimizes the use of user data and puts as much technology as possible on the mobile phone, not the cloud.
Metaio Computer Vision
In May 2015, Apple acquired Metaio, a technology company focused on computer vision and augmented reality. Metaio has more than ten years of research and development experience in computer vision, especially face recognition.
Polar Rose Face Recognition
In December 2011, Apple acquired Polar Rose, a Swedish face recognition company, for US$29 million. Polar Rose has launched several products, including FaceCloud, a facial recognition technology for web services, and the addition of functional FaceLib to mobile phones. Its face recognition software, Polar Rose can automatically circle the user's face through the face recognition.
Emollient face recognition
In January 2016, Emollient was included in the capsule of Apple. It can analyze human expression through face recognition technology to determine human emotions. Doctors can also use their company's technology to understand whether the patient's expression is painful, or they can apply this technique to monitors to see if anyone with "suspicious expression" is squatting in front of the product.
Flyby Media Computer Vision
Earlier this year, Apple quietly bought Flyby Media. Flyby Media has cutting-edge computer vision technology. In addition, its inertial sensing, simultaneous positioning and navigation space perception technology are also deeply loved by Apple. Flyby media has worked with Google to use computer vision technology to help Google develop Project Tango. Flyby Media helps the system to see and map its surroundings and is essential for driverless cars and augmented reality technology.
 Apple's AI principleUser Privacy> Data Training
Over the years, Apple has adhered to a principle that does not use user data for revenue. According to Apple officials, they are the only company that balances user privacy and user experience.
It is reported that Apple only crawls 200MB of cached data for machine learning calls on each user, including application habits, interaction with others, data for neural network processing, natural language model object recognition, face recognition, scene recognition, and so on. And these private information will not be uploaded to the network and cloud.
Related practitioners pointed out that if Apple continues to limit the use of user data, it will seriously hinder the research effect of Apple artificial intelligence. However, Apple believes that the obstruction of the data is insurmountable, and they will continue to adhere to this principle in the future.