Professors Hong-June Park and Youngjoo Lee of POSTECH Electrical Engineering have each had their teams selected for funding under the Samsung Future Technology Center’s 2017 designated theme, “Intelligence for Smart Machines.” Professor Seunggu Lee participates in Professor Youngjoo Lee’s team.

Professor Park’s project, “Real-Time Speech Recognition for Smartphones Using RNN/LSTM ASICs and an AP GPU,” stores the enormous number of machine learning parameters in DRAM and connects dedicated RNN/LSTM ASICs to DRAM through low-power, high-speed interfaces. Combining hardware and software with a smartphone GPU’s computing power, it is expected to implement low-power speech recognition operating locally without cloud computers, achieving over 95% accuracy and protecting privacy. Funding: KRW 900 million over three years.

Professor Youngjoo Lee’s project, “Autonomous Accuracy Scaling for Ultralow-Power Intelligent Mobile Systems,” with Professor Seunggu Lee, develops enabling technologies for implementing complex neural network algorithms on ultralow-power mobile platforms. It is expected to identify key requirements for commercializing and popularizing AI platforms. Funding: KRW 1.2 billion over three years.

Samsung Electronics operates its Future Technology Development Program with a total contribution of KRW 1.5 trillion over ten years beginning in 2013, supporting basic science, materials, ICT, and emerging and future technologies for national scientific development.