since dawn subordinate Iphone, many of the smart devices in smartphones came from elsewhere: the company’s computers known as Cloud. Mobile applications send user data to the cloud for useful tasks such as transcribing speech or suggesting responses to messages. right Now an Apple And the The Google Say smart phones They are smart enough to do some important and delicate things machine learning Such tasks themselves.
At the Apple WWDC event حدث This month, the company said its virtual assistant siri It will transcribe speech without clicking on the cloud in some languages on recent and future iPhones and iPads. while working I/O developer event last month, Google said the latest version of Android The operating system has a feature dedicated to securing the processing of sensitive data on the device, called Private Compute Core. Its primary uses include turning on the company’s smart reply feature that suggests responses to incoming emails and messages.
Both Apple and Google say on-device machine learning provides more privacy and faster apps. No transfer of personal data reduces exposure risks and saves time spent waiting for data to traverse the Internet. At the same time, data retention on devices is in line with the tech giants’ long-term interest in keeping consumers committed to their ecosystems. People who hear their data can be processed more privately may be more willing to agree to share more data.
The companies’ recent promotion of on-device machine learning comes after years of working on technology to restrict the data their clouds can “see”.
In 2014, Google began collecting some data about the use of the Chrome browser Through a technique called differential privacy, adding noise to the harvested data in ways that restrict what those samples reveal about individuals. Apple used this technology on data collected from phones to inform emoji, typing predictions, and web browsing data.
Recently, both companies have adopted a technology called Federal Learning. Allows cloud-based machine learning system to be updated without searching raw data; Instead, individual devices process the data locally and only share ingested updates. As with differential privacy, companies have discussed using federal learning only in limited cases. Google has used this technology to keep mobile typing predictions up to date with language trends; Apple has published research on its use in Update speech recognition models.
The rapid turnaround to doing some machine learning on phones has been amazing, says Rachel Cummings, an assistant professor at Columbia University who previously provided privacy advice to Apple. “It’s very rare to see something go from first concept to widespread in a very few years,” she says.
Not only did this advance require advances in computer science, but it also required that companies face the practical challenges of processing data on consumer-owned devices. Google said its federated learning system only taps users’ devices when they’re online, idle, and free online. This technology is enabled in part by improvements in the power of mobile processors.
Beefier mobile devices have also contributed to Google 2019 إعلان announcement The voice recognition of its virtual assistant on Pixel devices will be entirely on-device, free of a cloud crutch. Apple’s new on-device voice recognition technology for Siri, announced at WWDC this month, will use the company’s “neural engine” Added to its portable processors To enhance machine learning algorithms.
The technical feats are great. It is debatable to what extent they will purposefully alter users’ relationship with the tech giants.
Presenters at Apple’s WWDC said the new Siri design was a “major privacy update” that addressed the risks associated with accidentally transmitting voice to the cloud, saying that was users’ biggest privacy concern about voice assistants. Some Siri commands — such as setting timers — can be fully recognized locally, making for a quick response. However, in many cases, the commands that Siri copied – presumably from accidental recordings – will be sent to Apple’s servers to decode and respond to the programs. Siri’s cloud-based voice transcription will remain for HomePod smart speakers commonly installed in bedrooms and kitchens, where accidental recording can be even more alarming.
Google also promotes on-device data processing as a privacy gain and has indicated that it will expand this practice. The company expects partners like Samsung that use its Android operating system to adopt the new Privacy Compute Core and use it for features that rely on sensitive data.
Google has also made local analysis of browsing data a feature of its proposal for Reinventing Online Advertising Targeting, Dubbed FLoC He claimed to be more private. Academics and some rival tech companies have said the design is likely to help Google cement its dominance of online advertising by making targeting more difficult for other companies.