MISSION

Establishment of "usable" AI methodology

In artificial intelligence (AI), which has been actively introduced in various places, in order to maximize the performance of AI, the AI model needs to be optimized according to the operating environment (platform). Successes in various application cases are not the goal of our mission, but it's just a starting point to derive universal guidelines for designing "usable" AI.

Pursuit of next-generation AI

There are still various problems with AI, which is currently the most known. The problem we are focusing on is the problem of power consumption. AI models have become huge in recent years. To run such huge AI models in real time, a computer equipped with a GPU (graphic processing unit) commonly is used. Although the power saving of GPU processors has been achieved due to the progress of global development, there are still a lot of problems for portable use in terms of heat release problem due to long-term use and weight of drive battery. We will explore next-generation AI models that can fundamentally solve this problem.

Aids for Visually Impaired × AI

Up until now, infrastructure has been developed for the visually impaired, however, the circumstances of visually impaired are still not safe for going out with peace of mind. The SSC Lab is developing a device that supports those people to walk around streets safely and at ease.

Brain Machine Interface × AI

BMI is a technique to interface with computer by brain activities (EEG) signals. We are developing a methodology to measure the internal state of a person through EEG and applying it to a wide variety of applications, such as brain evaluations, diagnoses, BMI etc.

Pharmacology × AI

In clinically administering drugs, an adequate dose of drug to a patient is required for effectiveness and free from side effects. Responses to drugs vary from individual to individual. Blood is drawn to examine the reaction to the drugs. Frequent blood sampling puts a heavy burden on the patient. The frequency of blood sampling should be reduced as much as possible, and the appropriate amount of drug should be administered. We are developing an AI that makes an optimal dosing plan for each individual.

MEMBER

PROFILE