Koo Kwang-mo's AI Think Tank Achieves First Success... Development of 'Explainable AI' Technology

LG AI Research Institute Presents Two Papers at Prestigious International Conference
Joint Research with University of Toronto
Explainable AI Expected to Be Applied in Medical and Financial Fields
Continual Learning AI Reduces Memory Usage and Improves Performance by 40%

[Asia Economy Reporter Su-yeon Woo] LG Group's dedicated artificial intelligence (AI) technology organization, 'LG AI Research Institute,' has achieved its first results since its launch. With direct support from Koo Kwang-mo, Chairman of LG Group, the institute has begun to produce meaningful outcomes in the AI field, which is gaining attention as a new business area for LG Group, drawing interest from both inside and outside the industry.

On the 25th, LG AI Research Institute presented two papers in the fields of 'Explainable AI' and 'Continual Learning AI' at the Association for the Advancement of Artificial Intelligence (AAAI), the world's most prestigious AI conference. AAAI is a globally authoritative AI conference where leading AI research institutions worldwide present papers annually, serving as a benchmark for AI competitiveness across countries.

This research was conducted jointly with the University of Toronto in Canada, which LG co-established the 'Toronto AI Lab' with in 2018. It is also significant as the first achievement since LG elevated its existing AI-dedicated organization to an institute last year.

LG AI Research Institute's 'Explainable AI' refers to technology that not only provides results but also logically and easily explains the basis on which those results were derived. For example, after analyzing X-ray images, the AI not only identifies abnormalities in specific body parts but also provides detailed explanations of the reasoning behind the judgment. This technology is expected to be widely used in core AI development fields such as healthcare, finance, and law.

LG AI Research Institute conducted joint research with Professor Konstantinos Plataniotis's team at the University of Toronto, improving the accuracy and fidelity of explanations compared to existing methods and developing it for use in image recognition-related fields. LG explained that it is significant that a domestic company has achieved results in a field where even Google, a leader in AI, remains at the early research stage.

Another new technology, 'Continual Learning AI,' is a technique where AI learns data sequentially. Like humans, it uses short-term memory to store important past data and makes comprehensive judgments based on newly learned information.

Until now, a major challenge in AI learning was that learning large amounts of data caused a surge in memory usage, while reducing data volume led to decreased accuracy. The technology announced by LG AI Research Institute reduces the memory used during AI learning while improving learning performance by up to 40% compared to existing methods.

Koo Kwang-mo, Chairman of LG Group

LG Group is making comprehensive efforts to secure core AI technologies and build an ecosystem. Since 2018, it has been conducting joint research by establishing an AI lab with the University of Toronto in Canada, and in December last year, it officially launched the AI think tank 'AI Research Institute' and recruited Professor Honglak Lee, a world-renowned scholar from the University of Michigan, USA.

LG Group plans to expand the core AI research workforce to about 100 people by this year and operate customized training programs to nurture around 1,000 AI experts by 2023. It will invest 200 billion KRW over the next three years for talent acquisition and research and development.

Baek Kyung-hoon, head of LG AI Research Institute, said, "The goal of AI technology is to advance research so that customers can directly experience the development of technology," adding, "We will continuously recruit and nurture young AI talent to secure global top-level AI competitiveness."