iPhone 7 Unveiled: Apple's Deep Dive into AI and Digital Health

At 1:00 a.m. Beijing time on September 8, Apple held its annual fall launch event at the Bill Graham Civic Auditorium in San Francisco, USA. Apple released the iPhone 7/iPhone7 Plus、Apple Watch、AirPods Wireless Earbudsthe latest hardware products,In addition to these hardware and software products, Apple’s strategic layout in artificial intelligence appears easy to overlook. Previous commentary has suggested that Apple has fallen significantly behind Google, Facebook, and IBM in the field of AI, with no notable achievements beyond the well-known Siri, while its incrementally innovative hardware products seem to be gradually losing their appeal. In reality, artificial intelligence is playing an omnipresent “Brain"Role."

This August, Steven Levy, editor-in-chief of the prominent tech commentary outlet Backchannel, visited Apple and discovered that the company had already adopted neural network-based deep learning technology ahead of the industry. He subsequently published a major feature article titled “The iBrain is Here,” which detailed Apple’s ambitious plans in artificial intelligence.

In fact, Apple has already ported Siri’s speech recognition to a neural network-based system. The service was first made available to users in the United States and was rolled out globally on August 15. Some earlier technologies remain useful, including Hidden Markov Models (HMMs), but the current system employs machine learning techniques such as Deep Neural Networks (DNNs), Convolutional Neural Networks (CNNs), Long Short-Term Memory (LSTM) units, Gated Recurrent Units (GRUs), and n-grams. After the upgrade, Siri may appear unchanged on the surface, but it has been enhanced by deep learning. As with other underlying improvements, Apple has not disclosed details about Siri’s progress, preferring not to reveal its advancements to competitors. If users notice any difference, it is simply that Siri makes fewer errors. Apple also stated that the improvement in accuracy has been remarkable.

It is precisely because Apple designs its own chips that Apple engineers can collaborate directly with the chip design team responsible for writing firmware, thereby maximizing the performance of neural networks. The requirements of the Siri team have even influenced every aspect of iPhone design. When Apple’s neural network proves successful in one product, it can become the core technology for other products as well. Machine learning has enabled Siri to understand users and shifted input methods from manual typing to dictation. Thanks to Siri’s underlying technology, voice input has become smoother and more complete. In terms of natural language understanding, Siri began using machine learning to interpret user intent in November 2014 and launched a deep learning version a year later. As with speech recognition, machine learning has enhanced the user experience, particularly in understanding commands. In the iOS 10 update released on September 13 this year,SiriMajor upgrades are coming to image search, voice information, and more!

Machine learning is not just applied to Siri. Identifying unknown callers, listing your most frequently used apps after unlocking, flagging an appointment in Reminders (even if you haven’t added it to your calendar), and automatically displaying nearby tagged hotels—all these features can be refined to near perfection as Apple fully embraces machine learning and neural networks.

“Deep learning” is now ubiquitous across Apple’s products and services. The Apple Store employs deep learning to detect fraudulent warranty claims, while feedback from public beta versions of its operating systems is filtered using artificial intelligence to identify useful reports. Algorithms also enable automated review within the Health app. Additionally, Apple’s News app leverages machine learning to curate news sources likely to interest users.

Unlike companies such as Google, Facebook, and Microsoft, which have publicly established dedicated AI research institutions, Apple’s moves and acquisitions in the field of artificial intelligence appear to be aimed primarily at bolstering its existing or imminent business operations, with relatively less attention paid to long-term strategic planning—though it is also possible that Apple is simply keeping its cards close to its chest. As Apple itself has stated, “Apple occasionally acquires smaller technology companies, and we generally do not discuss our purpose or plans.””

Let’s briefly review Apple’s AI-related acquisitions over the past two years:

In March 2015, Apple acquired FoundationDB, a web-scale database technology company. Co-founded in 2009 by David Rosenthal, Nick Lavezzo, and Dave Scherer, the company’s product, FoundationDB, is a NoSQL database well-suited for cost-effective web applications.

In April 2015, Apple acquired LinX, an Israeli camera technology company, for approximately $20 million, according to The Wall Street Journal. The company designs camera modules featuring background bokeh, parallax imaging, and 3D image capture capabilities.

In May 2015, Apple acquired Coherent Navigation, a global positioning system (GPS) startup, and executives from the company joined Apple’s Maps team. Coherent Navigation’s primary research focus was on commercial high-precision navigation services based on satellite technology.

In September 2015, Apple quietly acquired Mapsense, a San Francisco-based startup specializing in map data analytics and visualization. Mapsense stated that its cloud-based, high-speed mapping system provides developers with critical data analytics and tools. Clients can upload terabytes of geotagged data to its service, after which the company generates customized visualizations using powerful search and filtering tools for developer analysis.

In the same month, Apple also acquired Faceshift, a Swiss company specializing in facial animation generation technology. Faceshift focused on real-time motion capture technology, with its core patent covering markerless facial motion capture. The company frequently collaborated with game and animation studios to achieve rapid and accurate facial expression capture using 3D sensors.

In October 2015, Apple acquired Vocal IQ, a UK-based natural language processing startup, likely to enhance its voice assistant, Siri. In the same month, Apple also acquired Perceptio, a startup whose technology enables enterprise clients to run advanced artificial intelligence systems on smartphones. Perceptio’s leaders, Nicholas Pinto and Zak Stone, are both renowned AI researchers specializing in image recognition systems based on deep learning technologies.

In January 2016, Apple acquired Emotient, an AI startup that leverages artificial intelligence to analyze facial expressions and interpret emotions.

That same month, Apple also confirmed its acquisition of the edtech startup LearnSprout. The San Francisco-based software company, founded three years prior, offers online data insights that help K-12 (preschool through secondary education) educators track student learning progress.

In August 2016, Apple acquired Turi, a machine learning and artificial intelligence startup. Turi enables developers to build applications equipped with machine learning and AI capabilities, as well as automated tuning features. Its product portfolio includes the Turi Machine Learning Platform, GraphLab Create, Turi Distributed, and Turi Predictive Services, primarily designed to help organizations of all sizes gain deeper insights from their data. Use cases include recommendation engines, fraud detection, customer churn prediction, sentiment analysis, and customer segmentation.

In addition, Apple announced in August that it had acquired the startup Gliimpse,Gliimpse was founded in Silicon Valley in 2013 by Anil Sethi and Karthik Hariharan, offering a unique service platform that enables users to consolidate medical and health data from various sources and share it with third parties (including physicians) as needed.The company’s business involves using machine learning technology to help people securely manage and share personal medical information. It is understood that the transaction was completed earlier this year, but Apple has not publicly disclosed it.

On the latest Apple Watch released this time, Apple has added a built-in GPS feature. With the second-generation Watch, it will accurately calculate your distance and cadence, and after your workout, a color-coded map will display your route and speed.All of this makes it highly suitable for use during outdoor activities such as running and mountain climbing.Users will benefit from an enhanced user interface, significantly improved performance, and new fitness and health features, including Activity Sharing.It also collaborated with NIKE to develop a customized version. Apple has long coveted the digital health sector.

The intelligent transformation of healthcare and wellness is another lucrative future market that has caught the eye of tech giants. Google, IBM, Microsoft, and Apple have all entered the arena, with Google and IBM in particular forging deep collaborations with healthcare institutions. At the 2014 Worldwide Developers Conference (WWDC), Apple unveiled its new health platform, HealthKit. Leveraging its hundreds of millions of iPhone users worldwide, Apple gains access to health data collected from smartphone and smartwatch sensors as well as a wide array of third-party accessories, forming the data foundation for its ambitions in the medical and healthcare sector.

This past May, Apple also hired Yoky Matsuoka, former Chief Technology Officer of Nest, to join its health products team. Yoky Matsuoka boasts an impressive background; she was a co-founder of Google X Lab and later joined the smart home company Nest in 2010 as its technology lead. Additionally, Apple has recruitedAnne Shelchuk from the ultrasound company Zonare, Craig Slyfield, an expert in 3D visualization of human bones, and Jay Mung, an expert in the wearables field who previously studied the sensor algorithms for Medtronic’s continuous glucose monitoring system.

On August 11 this year, Apple filed a new patent related to healthcare with the United States Patent and Trademark Office (USPTO). The patent reveals that Apple is developing a wearable medical device capable of rapidly measuring electrocardiograms (ECG) through a built-in array of sensors. By monitoring body movements, collecting raw data, and comparing it with previously stored data, the device enables more complex and precise data processing and analysis.Apple is indeed the most ideal company to enter the healthcare sector. However, Apple will not readily release a new product unless it has been fine-tuned to its optimal state. Perhaps soon, we will see Apple’s new products and major strategic moves in the healthcare field.