AI for mobile healthcare
Problem being addressed
The emergence and breakthrough of deep learning, that has been shown to achieve extraordinary results in a variety of real-world applications, such as skin lesion analysis, active authentication, facial recognition, botnet detection and community detection, is one of the primary drivers for mobile healthcare applications. However, since the deep learning techniques require enormous amount of computation, most of them cannot be directly deployed on the computation-constrained and energy-limited mobile and IoT devices.
An AI architecture for mobile healthcare systems. It enables the client to produce actionable intelligence locally using its embedded AI unit. When the satellite communication is available, the reduced feature data (or compressed raw data) could be uploaded to the server, and be processed by the networked AI units that utilize more powerful AI algorithms, thus generating more confident and detailed AI results.
Advantages of this solution
The experimental results show that the suggested framework is effective and efficient while switching the computation between embedded AI and networked AI. Also, our design of the framework’s decision unit consistently outperforms its baseline (i.e., randomly sending) in terms of both effectiveness and efficiency.
Solution originally applied in these industries
Possible New Application of the Work
Electronics and Sensors Industry
Deep learning techniques usually require enormous amount of computation, most of them cannot be directly deployed on the computation-constrained and energy-limited mobile and IoT devices. Optimizing the architecture for the smart home integration will allow more complex solutions, for both user's convenience and sustainability.
Source URL: #############