Samsung Advanced Institute of Technology (SAIT) recently presented a paper titled “Learning to Quantize Deep Networks by Optimizing Quantization Intervals With Task Loss” at this year’s Computer Vision and Pattern Recognition (CVPR), one of the top academic conferences in the field of computer vision, through which they introduced the company’s updated On-Device AI lightweight algorithm that can deliver optimization of low-power and high-speed computations.
On-Device AI technologies directly compute and process data from within the device itself which accounts for its enhanced performance as compared to existing algorithms. With the help of a Neural Processing Unit (NPU), AI can perform improved deep learning algorithm computation and process a heavy bulk of these with efficiency.
Samsung’s On-Device AI lightweight algorithm performs 8 times faster computations as compared to the existing 32-bit deep learning data used in servers.
The technology adjusts the data into groups of under 4 bits that help in maintaining accurate data recognition, through the ‘Quantization1 Interval Learning (QIL)’ process. This was successfully demonstrated by SAIT at the event.
This technology requires less hardware and less electricity which makes it easy to mount them inside the device.
Chang-Kyu Choi, Vice President and head of Computer Vision Lab of SAIT: “Ultimately, in the future, we will live in a world where all devices and sensor-based technologies are powered by AI. Samsung’s On-Device AI technologies are lower-power, higher-speed solutions for deep learning that will pave the way to this future. They are set to expand the memory, processor and sensor market, as well as other next-generation system semiconductor markets.”
On-Device AI technology directly computes data from within the device itself, operates on its own and provides quick and stable performance for use cases such as virtual reality and autonomous driving.
The benefits of On-Device AI technology include a reduction in the cost of cloud storage for AI operations, and safely saving onto the device personal biometric information used for device authentication such as fingerprint, iris and face scans.
Last month, Samsung Electronics had announced their goal of expanding their patented NPU technology and this recent On-Device AI lightweight algorithm solution is one step forward towards achieving that goal. It will also help develop Samsung’s system semiconductor capacity alongside strengthening On-Device AI processing, not only for mobile SoC but also for memory and sensor solutions in the near future.
