| ImpIx applies machine learning to its smart factory platform ‘OWP’ to enhance functionality

Impix (CEO Lee Sang-ho) and Soinet (co-CEOs Kim Yong-ho and Park Jeong-woo) announced on the 24th that they have jointly developed artificial intelligence that analyzes time series data from smart factories to predict quality, and have agreed to collaborate on an MLOps (Machine Learning Operations) project for smart factories.

The smart factory platform ‘OWP (One-Way Platform)’ developed by Impix is capable of collecting, storing, and analyzing production equipment sensor data, as well as generating and monitoring artificial intelligence learning models. Through the IoT device ‘N-series,’ it is possible to extract and collect data streams generated by PLCs and inspection/measurement sensors in production equipment, and process data based on standard protocol events. It can also be integrated with various industrial solutions based on big data system standard interfaces.

Impix, which has experience and expertise in building smart factories in various industrial fields, including large and medium-sized companies such as SK Hynix (Korea, China), Huons, Corens, Chong Kun Dang, and Korea Kolmar, is supplying the smart factory solution ‘OWP’ to enable companies to perform autonomous diagnosis and prescriptive response systems for production processes.

SoyNet recently secured investments from Enterprise Bank, Saltlux, and PlanH, and was selected as the 2020 Best Artificial Intelligence Company by CIO Advisor APAC. In addition, it has developed the ‘SoyNet’ AI inference-dedicated framework, which improves execution speed by three times and reduces memory usage to one-sixth compared to Google's TensorFlow, and has supplied it to domestic and international companies such as POSCO, NeuroMeka, Twim, Hyundai Steel, and SANfinity.

The collaboration between Impix and SoyNet, which applies real-time processing AI technology to massive manufacturing data processing, is expected to solve the speed and memory issues that have been obstacles to applying AI to smart factories. Furthermore, it is expected to accelerate the use of SoyNet's inference acceleration and lightweight framework in the ‘MLOps’ field, which is an adaptive AI service environment that retrains false positive and false negative data and reapplies it to the operating environment.

Soynet Vice President Ji Byung-jik emphasized, “Real-time AI application in edge environments is a unique advantage of Soynet's AI inference acceleration framework,” adding, “Soynet will become a leader in this field in the future.” Currently, Soynet is receiving support for commercialization and mentoring as an AI incubation company at the Gyeonggi Creative Economy Innovation Center.

Source: ZDNET Korea Reporter Bang Eun-ju