Journey of AI Innovation: LG AI Expert Course Achievement Sharing Session

LG AI Research operates specialized educational programs targeting employees across LG Group affiliates. The goal is to cultivate core AI talent capable of solving real-world problems within industrial settings. Central to this initiative is the 'LG AI Expert Course.'

This 12-week program features intensive 1:1 mentoring with AI experts from LG AI Research. Participants focus on resolving actual challenges faced in their current roles by utilizing cutting-edge AI technology, thereby maximizing the improvement of practical problem-solving skills and the potential for real-world application.

LG AI Expert Course: Top 3 Outstanding Projects

The recent LG AI Expert Course ran for 12 weeks starting in July. 11 participants, guided by their mentors, engaged in comprehensive processes including problem definition, exploration of the latest papers and technologies, AI model development, and establishing application plans for their respective fields. On November 27th, a meaningful ceremony and achievement sharing session were held to celebrate the successful completion of the course.

The event was honored by the presence of Woohyung Lim and Honglak Lee, co-head of LG AI Research, along with other key leaders, who attended to congratulate the ceremony. At this gathering, 11 participants demonstrated their capabilities by presenting research results across various domains, including Vision and Language.

Project evaluations were conducted by combining scores from a judging panel (consisting of mentors who assessed technical completeness) and a company-wide employee preference vote. This process resulted in the selection of the top three projects, validating LG's AI capabilities.

1st Place: Jiyoung Kim (LG Display) | Mentor: Moonjung Eo (Data Intelligence Lab)

The first-place achievement was secured by Jiyoung Kim from LG Display for his research on 'Cavity Prediction utilizing Visibility Spectrum.' Collaborating with mentor Moonjung Eo, she applied an AI model to the crucial process of diagnosing and correcting cavities essential during OLED device structure development. He developed a deep learning model that ensures high prediction accuracy and reliable prediction scatter for cavities, significantly improving the time and cost efficiency of the development process. Kim shared his thoughts on concluding the program: "I gained significant insights by completing what I thought was a difficult project in a short 12-week period. Thanks to Mentor Moonjung Eo, I was able to dedicate time to scholarly consideration of the model architecture."

Moonjung Eo, mentor at the Data Intelligence Lab, offered her remarks: “Thanks to Jiyoung’s passion and initiative, I was able to dedicate more thought to the project and provide stronger support. I truly congratulate him on the first-place win and hope he continues to successfully apply AI in many areas within his field.”

2nd Place: Jungsun Im (LG Electronics) | Mentor: Jinhyung Kim (Vision Lab)

The second prize went to Jungsun Im of LG Electronics for her research focused on improving the performance of an age-reasoning model in a real-world service operating environment. Working with Jinhyung Kim from the Vision Lab, she concentrated on enhancing the practical reasoning performance of the age-reasoning model used in the signage advertisement optimization service (based on floating population characteristics). She built a dataset similar to the real environment using VLM (Vision Language Model) and clustering techniques, then applied post-processing utilizing non-facial information to boost age- reasoning performance.

Lim remarked upon completing the program: "I was able to generate many great ideas with Jinhyung Kim, and thanks to him, I successfully finished this excellent course.".

3rd Place: Gyudong Sim (LG Electronics) | Mentor: Minjun Park (Language Lab)

The third prize was awarded to Gyudong Sim of LG Electronics. His research, titled 'May AI help you_Data Extraction Code Generation Agent,' aimed at streamlining data extraction tasks. He applied a Text-to-SQL Agentic approach to convert natural language data requests submitted through the JIRA system into corresponding SQL queries. Ultimately, he developed an Agent that generates the necessary data extraction code in Jupyter Notebook format and provides it to the responsible engineer.

He shared his feelings about the program, saying, "Thanks to the mentoring from Minjun Park of the LG AI Research Language Lab, I was able to clarify the project's direction and bring it to completion. It was an opportunity for great growth through collaborative brainstorming."

The outstanding projects from this LG AI Expert Course—spanning Data Intelligence, Vision, and Language—demonstrate that LG's AI capabilities can deliver tangible results across various industrial domains, including manufacturing, visual processing, and natural language processing.

Moving forward, LG AI Research will continue to provide diverse AI education, including the Expert Course, aiming to solve real-world industry challenges. The knowledge gained and problem-solving experience acquired through these processes will be disseminated throughout the LG Group, ensuring that its core personnel possess the necessary expertise to lead future AI projects within the organization.