SUNNYVALE, Calif., April 18, 2017 –Socionext Inc. and SOINN Inc. today announced initial results of collaboration started in 2016, in which Socionext extracts and delivers biometrics data to the "Artificial Brain SOINN". The companies achieved initial results in reading ultrasound images from Socionext's viewphii™ mobile ultrasound solution by Artificial Brain SOINN. The results will be introduced at Medtec Japan, held in Tokyo Big Sight, April 19-21, at booths 4505 & 4507.
In this initial trial, SOINN learned to read subcutaneous fat thickness from abdominal ultrasound images. The estimations by SOINN were then compared with the reading results by ultrasound technicians. Machine deep leaning, which is attracting more and more attentions recently, is thought to require hundreds of thousands of images in order to learn from reading the images. In contrast, SOINN needed only about 700 images, revealing the unique approach by SOINN Inc. to obtain learning result using very small data samples. SOINN can accurately read fat tissue thickness from 80 percent of the data within 5 percent margin of error. There were noticeable differences after reviewing the data group from the readings by human and by SOINN. After reviewing these data, it was confirmed that human error, including numerical input, was a common occurrence from data reading by human. Based on the findings, the companies believe that AI has the potential to be used for assisting technicians in reading images and detecting human errors in medical image handling.
Socionext will build on the knowledge gained by the current trials and continue to develop technology and business promotion on the use of AI, not only in the medical field but for various IoT applications. Socionext has its core strength in extracting value-added information from sensor data, as well as the ability to develop edge computing solutions built on the experience of developing high-performance, multi-functional processors. The company will leverage such expertise and apply innovative methods of transferring sensor data to the cloud, to make full use of such data in various applications. By achieving this goal, it aims to become an industry leader in the use of AI for IoT.
SOINN Inc. is a unique startup founded by Associate Professor of Tokyo Institute of Technology which is one of the top computer science universities in Japan. We drive a safe and efficient world with our innovative Machine Learning technologies. Our patent based, Artificial Brain "SOINN", enables machines to be autonomous and smart. Our cutting edge knowledge and skills get over many kinds of tasks that Deep Learning cannot achieve. SOINN offers low CPU power, shorter learning curve and time to market for achieving a smart world with IoT.
SOINN = Self-Organizing Incremental Neural Network
About Socionext America Inc.
Socionext America Inc. (SNA) is the US branch of Socionext Inc., one of the world’s leading fabless SoC suppliers. Headquartered in Milpitas, California, the company provides leading-edge technologies and offers a wide range of standard and customizable SoC solutions. Socionext America meets customers' requirements with quality semiconductor products based on extensive and differentiated IPs, proven design methodologies, and state-of-the-art implementation expertise, with full support.
About Socionext Inc.
Socionext Inc. is a global SoC (System-on-Chip) supplier and a pioneer of a unique “Solution SoC” business model through decades of industry experience and expertise. Socionext contributes to global innovation in advanced technologies including automotive, data center, networking, and smart devices. As a trusted silicon partner, Socionext delivers superior features, performance, and quality that differentiate its customers’ products and services from their competition.
Socionext Inc. is headquartered in Yokohama, and has offices in Japan, Asia, United States and Europe to lead its development and sales activities. For more information, visit https://www.socionext.com/en/.
All company or product names mentioned herein are trademarks or registered trademarks of their respective owners. Information provided in this press release is accurate at time of publication and is subject to change without advance notice.