LG AI Expert Course 2023 First Graduation & Achievements Sharing Session

As the application of AI continues to extend across various industries, the importance of AI-related knowledge and capabilities is growing, not just among developers but also within diverse job functions. In response to this evolving environment, LG AI Research is running a variety of programs aimed at enhancing the AI competence of LG Group members.

Among them, the LG Expert Course is an educational program where AI specialists from LG AI Research mentor students one-on-one for 12 weeks, tackling actual operational tasks of affiliate companies with AI technology to strengthen AI proficiency. This program is conducted biannually in the first and second halves of the year.

The first phase of the 2023 Expert Course ran from February 14 to May 4. Participants included 13 trainees from four affiliates - LG Display, LG Chem, LG Energy Solution, and LG Electronics, mentored by AI Specialists from the Applied AI Research, Materials Informatics, Vision, and Language departments of LG AI Research. A completion ceremony and achievements presentation session to celebrate the successful conclusion and share the results accomplished over the 12 weeks was held on June 30 at the Vision Hall in the LG Science Park ISC building.

In an era where AI is being integrated into every field, the essential virtue needed is curiosity.

The completion ceremony began with a rousing speech from Jungyun Seo, the AI Talent Development Supervisor of LG AI Research Supervisor Seo initially expressed his appreciation and offered encouragement to the graduates and mentors who had tirelessly collaborated to address various issues over the course of the 12 weeks. He then reflected on the contemporary climate, where AI technology is being integrated across a multitude of domains.

In his speech, he noted, “Given that AI technology is now permeating every sector, it’s become a challenging world for those unable to develop new technologies to lead the market. I hope that through the LG AI Expert Course, different LG affiliates will identify issues they can address with AI to yield results, fostering a synergistic relationship with LG AI Research.”

He underscored the virtue of curiosity for creative integrative talents, inspiring them with, “If you continually harbor curiosity and ponder on how AI can be incorporated into everything around you, you’ll encounter innovative and highly beneficial ideas that you may have overlooked.”

Ceremony for the presentation of certificates and awards for outstanding projects recognizing 12 weeks of accomplishments

Following the speech, Kyunghoon Bae, Chief of LG AI Research, presented graduation certificates to the 13 trainees who completed the first term of the 2023 LG AI Expert Course. The event continued with the Best Practice award ceremony, acknowledging three distinguished projects.

First place was awarded to Joohyup Kim from LG Electronics for “Strategic Agenda Trend Sensing.” Second place went to Seulchan Park from LG Chem for “Synthesizing Potential OLED Materials,” and third place was secured by Jongguk Park from LG Electronics for “3D Object Recognition Technology as Perceived by Robots.” Outstanding projects were selected based on internal evaluations considering factors like business impact, technical challenges, and potential for educational growth, along with feedback from online employee voting.

Chief Bae, during his congratulatory remarks, stated, “I firmly believe that AI technology will offer significant aid in various aspects of our work. If you can effectively utilize AI technology, it will greatly contribute to personal growth. I encourage you all to maintain interest and continuously strengthen your capabilities in AI technology.”

Sharing of Accomplishments and Vision Through Excellent Project Presentations

Following the awards ceremony, the highlight of the event, the achievements presentation, took place. The top three winners detailed their research backgrounds, goals, processes, outcomes, improvement plans, and implementation strategies in their businesses, as well as sharing their personal reflections.

BP Third Place Research: Development of a Real-time, High-performance 3D Object Recognition System for Indoor Autonomous Robots’ Spatial Understanding (Spatial AI) - Jongguk Park, LG Electronics

Jongguk Park from LG Electronics, in collaboration with his mentor Bumsoo Kim from LG AI Research’s Vision Lab, conducted research on “Developing a Real-time, High-performance 3D Object Recognition System for Indoor Autonomous Robots’ Spatial Understanding (Spatial AI).” The objective of this project was to develop a deep learning model to enhance a robot’s 3D object recognition capabilities and inference speed.

While our real world is 3D, most tasks and data are processed into 2D information, creating an information asymmetry with the real world. This results in a limitation when visualizing objects with 3D information only in 2D, as it is not intuitive. Due to these issues, recent research on 3D detectors has been active but applying it in real-time applications remains challenging.

In response to this, we experimented with and evaluated various models to develop a deep learning model that can detect 3D information of objects using RGB-D sensor data without decreasing recognition performance while improving inference speed.

After his presentation, Jongguk Park shared, “Thanks to the timely and effective coaching from my mentor, I was able to achieve the project objectives without wasting time. Additionally, discussing our individual projects with other participants and enhancing our ideas weekly was a fruitful process.”

BP Second-place Research Award Recipient: Synthesis of Viable OLED Materials - Seulchan Park, LG Chem

Next, Seulchan Park from LG Chem, who secured the second place, continued with his presentation. The project, “Creating Synthesizable OLED Materials,” mentored by Sungmoon Ko from the Materials Informatics Lab at LG AI Research, aims to convert OLED materials and unknown compounds into synthesizable molecules using deep learning technology and to propose synthetic pathways.

While there’s vibrant ongoing research into the creation of molecules using deep learning technology and the production of various molecules with desired characteristics, synthesizing these computer-designed molecules in a laboratory presents a significant challenge.

To mitigate this, Seulchan Park researched a method to generate synthesizable molecules and synthetic pathways using the Conditional Reaction Sequence Variational Autoencoder (C-RSVAE) model. The C-RSVAE model, previously presented in a paper by LG AI Research, has garnered considerable attention.

Seulchan Park commented, “The process of using deep learning models to solve problems was innovative. It enhanced my understanding of molecule generation models and reemphasized the importance of quality data.”

BP First-place research award recipient: Strategic Agenda Trend Sensing - Joohyup Kim, LG Electronics

Joohyup Kim’s “Strategic Agenda Trend Sensing” project was highly praised for reflecting persistent needs in strategic and new business areas, and its potential for immediate application in current operations. Minseok Cho from the Language Lab of LG AI Research acted as a mentor for this project.

Joohyup Kim explained the project’s background, “To promptly respond to rapidly changing external environments and shifts in customer purchasing behaviors, it’s crucial to quickly identify and address relevant business news.” To achieve this, the project aimed to develop a preemptive trend sensing solution using deep learning for collecting and classifying decision-making data, rather than relying on humans.

The project used four pre-training models from the Bidirectional Encoder Representations from Transformers (BERT) series for fine-tuning, and the most accurate model was selected. The selected model was implemented in several messenger programs and drew a lot of attention when demonstrated.

Joohyup Kim shared, “Although I was initially concerned due to my background and work areas being distantly related to AI, I learned a great deal thanks to my mentor. Despite the short duration, this was a precious opportunity to study artificial intelligence in earnest.”

After the BP project presentations, mentors from LG AI Research, Minseok Cho and Bumsoo Kim, shared their reflections. Minseok Cho said, “It was enjoyable to discuss interesting topics every week and observe the progress of the passionate participants. I am ready to provide assistance whenever necessary in the future.”

Bumsoo Kim expressed his pride, saying, “Object Detection is a difficult field to navigate due to its extensive code base, even for those specialized in AI. The amount of effort put into unseen areas was commendable. I’m genuinely proud of the positive outcomes.”

As the LG AI Expert Course progresses, even departments with little direct association with AI are increasingly incorporating AI into their problem-solving efforts. Participants of the Expert Course are expected to apply the latest AI models taught by LG AI Research to their real-world problems, learning how to implement AI models and spreading the educational benefits among their peers. LG AI Research plans to continue running diverse programs to nurture AI talents across various fields in the future.