May 7, 2019 at 09:57Machine Learning in Mobile Development: Prospects and Decentralization
- Transfer
Good morning, Habr!
We have nothing to add to the heading of the article in our notice - therefore, everyone is immediately invited to cat. We read and comment. Developers for mobile devices will benefit from the revolutionary changes that machine learning on devices can offer today . The point is how much this technology enhances any mobile application, namely, it provides a new level of convenience for users and allows you to actively use powerful features, for example, provide the most accurate recommendations based on geolocation , or instantly detect diseases in plants .
Such a rapid development of mobile machine learning is the answer to a number of common problems that we had time to learn in classical machine learning. In fact, everything is obvious. In the future, mobile applications will require faster data processing and a further reduction in latency.
You might have wondered why AI-based mobile apps, they can’t simply run inference in the cloud. Firstly, cloud technologies depend on central nodes (imagine a huge data center where both extensive data warehouses and large computing power are concentrated). With such a centralized approach, it is impossible to cope with the processing speeds sufficient to create smooth mobile interactions based on machine learning. Data should be processed centrally, and then sent back to the device. This approach takes time, money and does not guarantee the privacy of the data itself.
So, having outlined these main advantages of mobile machine learning, let us examine in more detail why the revolution in machine learning unfolding before our eyes should be interesting to you personally as a mobile developer.
Mobile application developers know that increased latency can become a black mark for a program, regardless of how good its features are or how respectable the brand is. Earlier on Android devices, there were serious delays in many video applications , because of which viewing video and audio often turned out to be out of sync. Similarly, a high latency social media client can turn communication into a real torture for the user.
Implementing machine learning on a device is becoming increasingly important precisely because of such problems with delays. Imagine how image filters for social networks work, or restaurant recommendations with reference to geolocation. In such applications, the delay should be minimal, only in this case it can work at the highest level.
As mentioned above, cloud processing is sometimes slow, and the developer needs the delay to tend to zero - only in this case the machine learning capabilities in the mobile application will work as it should. Machine learning on devices opens up such data processing capabilities that really allow reducing the delay to almost zero.
Smartphone makers and tech giants are slowly starting to realize this. For a long time, Apple remained the leader in this industry, developing more and more advanced chips for smartphones using its Bionic system, which implements the neural engine Neural Engine, which helps to drive neural networks directly on the device, while achieving incredible speeds .
Apple also continues to step by step develop Core ML, its machine learning platform for mobile applications; in the TensorFlow Lite libraryAdded support for GPUs; Google continues to add preloaded features to its ML Kit machine learning platform. It is with these technologies that you can develop applications that allow you to instantly process data, eliminate any delays and reduce the number of errors.
This combination of accuracy and seamless user interactions is a key indicator that mobile app developers should consider when incorporating machine learning capabilities into them. And in order to guarantee such functionality, it is required to adopt machine learning on devices .
Another huge benefit of edge computing that cannot be overestimated is how much they improve the security and privacy of users. The guarantee of data security and privacy in the application is an integral part of the developer’s tasks, especially taking into account the need to comply with the GDPR (General Data Protection Regulation), the new European laws, which will undoubtedly affect the practice of mobile development.
Since the data does not need to be sent for processing to the north or to the cloud, cybercriminals have fewer opportunities to exploit any vulnerabilities that occurred during such a transfer; therefore data integrity is maintained. This makes it easier for mobile application developers to comply with GDPR data security regulations.
Machine learning on devices also provides decentralization, largely on the same principle as blockchain. In other words, it is more difficult for hackers to put a connected network of hidden devices with a DDoS attack than to carry out the same attack on a central server. This technology can also be useful when working with drones and to monitor compliance with the law.
The aforementioned smartphone chips from Apple also contribute to improving the security and privacy of the user - so they can serve as the basis for Face ID. This iPhone feature is based on a neural network deployed on devices and collecting data about all the various representations of the user’s face. Thus, the technology serves as an extremely accurate and reliable method of identification.
Such and newer AI-enabled equipment will pave the way for safer user interactions with the smartphone. In fact, developers get an extra layer of encryption to protect user data.
Apart from the latency issues, sending data to the cloud for processing and extracting leads requires a good Internet connection. Often, especially in developed countries, there is no need to complain about the Internet. But what to do in areas where communication is worse? When machine learning is implemented on devices, neural networks live on phones on their own. Thus, the developer can deploy the technology on any device and in any place, regardless of the quality of the connection. Plus, this approach leads to the democratization of ML capabilities .
Healthcare- One of the industries that can especially benefit from machine learning on devices, as developers can create tools that check vital indicators, or even provide robosurgery without any Internet connection. This technology is also useful for students who want to access lecture materials without having an Internet connection - for example, being in a transport tunnel.
Ultimately, machine learning on devices will provide developers with tools to create tools that will be useful to users from all over the world, regardless of the situation with the Internet connection. Given that the power of new smartphones will be at least no lower than current ones, users will forget about problems with delays when they work with the application offline.
Machine learning on devices is also designed to save you a fortune - because with it you will not have to pay external contractors who would implement and support many solutions. As mentioned above, in many cases you can do without the cloud, and without the Internet.
GPUs and AI-specific cloud services are the most expensive solutions you can purchase. When launching models on the device, you do not have to pay for all these clusters, due to the fact that today more and more advanced smartphones equipped with neuromorphic processors (NPU) are appearing .
By avoiding the nightmare of heavy data processing between the device and the cloud, you save tremendously; therefore, implementing machine learning solutions on devices is very beneficial. In addition, you save because the data bandwidth requirements are significantly reduced in your application.
Engineers themselves also greatly save on the development process, since they do not have to collect and maintain additional cloud infrastructure. On the contrary, it is possible to achieve more with the forces of a smaller team. Thus, human resource planning in development teams is much more efficient.
Undoubtedly, in the 2010s, clouds became a real blessing that simplified data processing. But high technology is developing exponentially, and machine learning on devices could soon become the de facto standard not only in the field of mobile development, but also in the field of the Internet of Things.
Due to reduced latency, improved security, offline capabilities and overall cheaper costs, it is not surprising that the largest mobile development players are betting on this technology. Mobile application developers should also take a closer look at it to keep up with the times.
We have nothing to add to the heading of the article in our notice - therefore, everyone is immediately invited to cat. We read and comment. Developers for mobile devices will benefit from the revolutionary changes that machine learning on devices can offer today . The point is how much this technology enhances any mobile application, namely, it provides a new level of convenience for users and allows you to actively use powerful features, for example, provide the most accurate recommendations based on geolocation , or instantly detect diseases in plants .
Such a rapid development of mobile machine learning is the answer to a number of common problems that we had time to learn in classical machine learning. In fact, everything is obvious. In the future, mobile applications will require faster data processing and a further reduction in latency.
You might have wondered why AI-based mobile apps, they can’t simply run inference in the cloud. Firstly, cloud technologies depend on central nodes (imagine a huge data center where both extensive data warehouses and large computing power are concentrated). With such a centralized approach, it is impossible to cope with the processing speeds sufficient to create smooth mobile interactions based on machine learning. Data should be processed centrally, and then sent back to the device. This approach takes time, money and does not guarantee the privacy of the data itself.
So, having outlined these main advantages of mobile machine learning, let us examine in more detail why the revolution in machine learning unfolding before our eyes should be interesting to you personally as a mobile developer.
Delay reduction
Mobile application developers know that increased latency can become a black mark for a program, regardless of how good its features are or how respectable the brand is. Earlier on Android devices, there were serious delays in many video applications , because of which viewing video and audio often turned out to be out of sync. Similarly, a high latency social media client can turn communication into a real torture for the user.
Implementing machine learning on a device is becoming increasingly important precisely because of such problems with delays. Imagine how image filters for social networks work, or restaurant recommendations with reference to geolocation. In such applications, the delay should be minimal, only in this case it can work at the highest level.
As mentioned above, cloud processing is sometimes slow, and the developer needs the delay to tend to zero - only in this case the machine learning capabilities in the mobile application will work as it should. Machine learning on devices opens up such data processing capabilities that really allow reducing the delay to almost zero.
Smartphone makers and tech giants are slowly starting to realize this. For a long time, Apple remained the leader in this industry, developing more and more advanced chips for smartphones using its Bionic system, which implements the neural engine Neural Engine, which helps to drive neural networks directly on the device, while achieving incredible speeds .
Apple also continues to step by step develop Core ML, its machine learning platform for mobile applications; in the TensorFlow Lite libraryAdded support for GPUs; Google continues to add preloaded features to its ML Kit machine learning platform. It is with these technologies that you can develop applications that allow you to instantly process data, eliminate any delays and reduce the number of errors.
This combination of accuracy and seamless user interactions is a key indicator that mobile app developers should consider when incorporating machine learning capabilities into them. And in order to guarantee such functionality, it is required to adopt machine learning on devices .
Enhanced Security and Privacy
Another huge benefit of edge computing that cannot be overestimated is how much they improve the security and privacy of users. The guarantee of data security and privacy in the application is an integral part of the developer’s tasks, especially taking into account the need to comply with the GDPR (General Data Protection Regulation), the new European laws, which will undoubtedly affect the practice of mobile development.
Since the data does not need to be sent for processing to the north or to the cloud, cybercriminals have fewer opportunities to exploit any vulnerabilities that occurred during such a transfer; therefore data integrity is maintained. This makes it easier for mobile application developers to comply with GDPR data security regulations.
Machine learning on devices also provides decentralization, largely on the same principle as blockchain. In other words, it is more difficult for hackers to put a connected network of hidden devices with a DDoS attack than to carry out the same attack on a central server. This technology can also be useful when working with drones and to monitor compliance with the law.
The aforementioned smartphone chips from Apple also contribute to improving the security and privacy of the user - so they can serve as the basis for Face ID. This iPhone feature is based on a neural network deployed on devices and collecting data about all the various representations of the user’s face. Thus, the technology serves as an extremely accurate and reliable method of identification.
Such and newer AI-enabled equipment will pave the way for safer user interactions with the smartphone. In fact, developers get an extra layer of encryption to protect user data.
No internet connection required
Apart from the latency issues, sending data to the cloud for processing and extracting leads requires a good Internet connection. Often, especially in developed countries, there is no need to complain about the Internet. But what to do in areas where communication is worse? When machine learning is implemented on devices, neural networks live on phones on their own. Thus, the developer can deploy the technology on any device and in any place, regardless of the quality of the connection. Plus, this approach leads to the democratization of ML capabilities .
Healthcare- One of the industries that can especially benefit from machine learning on devices, as developers can create tools that check vital indicators, or even provide robosurgery without any Internet connection. This technology is also useful for students who want to access lecture materials without having an Internet connection - for example, being in a transport tunnel.
Ultimately, machine learning on devices will provide developers with tools to create tools that will be useful to users from all over the world, regardless of the situation with the Internet connection. Given that the power of new smartphones will be at least no lower than current ones, users will forget about problems with delays when they work with the application offline.
Reduce costs for your business
Machine learning on devices is also designed to save you a fortune - because with it you will not have to pay external contractors who would implement and support many solutions. As mentioned above, in many cases you can do without the cloud, and without the Internet.
GPUs and AI-specific cloud services are the most expensive solutions you can purchase. When launching models on the device, you do not have to pay for all these clusters, due to the fact that today more and more advanced smartphones equipped with neuromorphic processors (NPU) are appearing .
By avoiding the nightmare of heavy data processing between the device and the cloud, you save tremendously; therefore, implementing machine learning solutions on devices is very beneficial. In addition, you save because the data bandwidth requirements are significantly reduced in your application.
Engineers themselves also greatly save on the development process, since they do not have to collect and maintain additional cloud infrastructure. On the contrary, it is possible to achieve more with the forces of a smaller team. Thus, human resource planning in development teams is much more efficient.
Conclusion
Undoubtedly, in the 2010s, clouds became a real blessing that simplified data processing. But high technology is developing exponentially, and machine learning on devices could soon become the de facto standard not only in the field of mobile development, but also in the field of the Internet of Things.
Due to reduced latency, improved security, offline capabilities and overall cheaper costs, it is not surprising that the largest mobile development players are betting on this technology. Mobile application developers should also take a closer look at it to keep up with the times.