Samsung Is Stepping Towards Artificial Intelligence

Samsung Is Stepping Towards Artificial Intelligence

As Samsung is pushing further with the creation of its proprietary Neural Processing Units, the company has created a brand new high-speed, low-power On-Device AI deep learning algorithm promising 4 times more power efficiency and 8 times more efficiency than existing solutions. The Samsung Advanced Institute of Technology detailed the new On-Device AI technology in a current paper claiming that the algorithm has been efficiently examined. It should make way for a future in which “AI powers all devices and sensor-based technologies.”

This increase in efficiency and energy efficiency has been achieved “by adjusting the data into teams of under 4 bits whereas maintaining accurate data recognition,” or what the corporate calls Quantization Interval Learning (QIL). Computations performed utilizing the QIL process can achieve outcomes similar to existing options while requiring 1/40 to 1/120 transistors.

This technique of compressing data into bit groups of 4 bits or lower permits for thousands of deep studying computations to be processed at the same time. It additionally allows for “and” “or” computations to be performed along with arithmetic calculations such as multiplication.

With a faster, much less demanding On-Device solution now being within the works, we will only count on that features such as pictures and AR will turn out to be smarter, faster, and more accurate with the brand new generation of devices.

However, Samsung won’t stop at cellular chip-sets. The corporate also revealed that it wants to apply this new deep learning algorithm to memory and sensor solutions within the close to future. “We’ll live in a world where all gadgets and sensor-based technologies are powered by AI” in accordance with the VP of Computer Vision Lab of Samsung Advanced Institute Technology, Chang-Kyu Choi. And the future is based on embracing On-Device AI technologies slightly than relying too much on cloud-based AI solutions.

Leave a Reply

Your email address will not be published. Required fields are marked *