Master’s candidate Taesoo Kim from Professor Jeongho Kim’s lab achieved remarkable success by winning the Best Paper Award at the ‘IEEE Electrical Design of Advanced Packaging and Systems (EDAPS) 2024‘, held on December 17-19 at the Taj Yeshwantpur, in Bangalore, India.
The ‘IEEE EDAPS’ is the largest and most influential conference on semiconductor packaging technology in the Asia-Pacific region. Since 2002, it has been annually organized and hosted by the IEEE Electronic Packaging Society.
< Taesoo Kim presenting at the EDAPS >
The award-winning paper, titled “Design and Analysis of Twin Tower High Bandwidth Memory (HBM) Architecture for Large Memory Capacity and High Bandwidth System,” was authored by mater’s student Taesoo Kim as the first author.
This paper received high acclaim for its creativity, completeness and future scalability. In particular, the innovative concept of implementing the “Twin Tower” structure to effectively address the limitations of traditional memory designs, and the ability to present a new direction in memory architecture for supporting ultra-large AI models, were recognized and led to the prestigious Best Paper Award.
< Professor Euijong Whang Elected as a New Y-KAST Member >
Professor Euijong Whang from our department has been elected as a member of the Young Korean Academy of Science and Technology (Y-KAST) of the Korean Academy of Science and Technology (KAST) for 2025.
Launched in February 2017, Y-KAST is the only young academy in Korea, consisting of outstanding young scientists under the age of 45 who engage in policy activities and international exchanges.
Y-KAST members are selected from young scientists under the age of 43 who have demonstrated outstanding academic achievements. In particular, the selection process focuses on achievements made as independent researchers in Korea after obtaining a Ph.D., ultimately choosing next-generation science and technology leaders with the potential to contribute significantly to the advancement of science and technology in Korea.
<Prof. Euijong Whang (third from left) attending the Y-KAST Member’s Day event held on December 19.>
Professor Euijong Whang has been recognized as a young scientist conducting pioneering research in the fields of big data and artificial intelligence (AI). He has led efforts in the area of data-driven responsible AI research, collaborated with global IT companies, and became the first in Asia to receive the Google AI Award. His outstanding achievements on the global stage led to his selection as a Y-KAST member in the engineering division.
With this, our department now proudly counts a total of five current Y-KAST members, including Professors Euijong Whang, Hyunjoo Lee, Junil Choi, Minsoo Rhu, and Min Seok Jang. Additionally, two alumni, Professors Joonwoo Bae and Changho Suh, have previously served as Y-KAST members.
<(From the left) Professors Kyung Cheol Choi and Jun Kyun Choi>
The research achievements of Professors Kyung Cheol Choi and Jun Kyun Choi from our department have been selected for the ‘2024 National R&D Excellence 100,’ organized by the Ministry of Science and ICT.
The National R&D Excellence 100 is a program established to enhance public understanding and interest in the role of science and technology in driving national development and to instill pride among scientists and engineers. Since 2006, outstanding national R&D achievements have been selected across ministries.
This year, a total of 869 achievements recommended by ministries, agencies, and offices were nominated. Following evaluations by a selection committee consisting of 100 industry-academic-research experts and public verification, the final 100 outstanding achievements were chosen.
<Fiber OLED that stably emits light under bending conditions >
In the field of Information/Electronics, Professor Kyung Cheol Choi’s achievement, ‘Development of the World’s Longest-Lifespan Fiber-Based OLED for Wearable Displays,’ was recognized. This technology was designed to maximize user comfort and convenience by creating fiber-type displays.
The fiber organic light-emitting diode (OLED) developed by Professor Kyung Cheol Choi is designed to operate on a single cylindrical thread. It can be produced using existing thermal deposition equipment employed in current industries, enhancing its industrial applicability and compatibility.
Notably, the fiber OLED operates stably for up to 720 hours, which is nine times longer than the previously reported maximum lifespan of 80 hours. This achievement is recognized as the highest-performing technology globally.
The light-emitting device implemented on a single thread, along with the wearable display that will be developed based on it, contributes not only to the fiberization of light-emitting devices but also sensors, batteries, and various electronic components. This core technology is expected to drive the development of diverse wearable electronic textiles. It is anticipated to spark innovation in medical and safety industries through wearable clothing-type displays and to create ripple effects across various scientific and industrial fields.
< ITU-T International Standard Adoption >
Another outstanding achievement in the Information/Electronics category is Professor Jun Kyun Choi’s ‘Development of IoT Trust Enabler Technology and Leadership in International Standards.’ This technology was developed to establish a reliable and secure intelligent IoT ecosystem.
The trust modeling algorithm developed by Professor Jun Kyun Choi for building IoT ecosystems secured approval for three international standards (Y.3057, Y.3058, Y.3060) related to IoT data trust frameworks and trust-based service delivery structures at ITU-T, the top international organization in the field of information and communication technology.
Of the 49 international standard contributions submitted, 38 were adopted, demonstrating a high adoption rate and highlighting his leadership in global standardization.
To support these standardization efforts, the trust analysis and IoT device operation technologies developed were published in 17 papers, including more than 10 in top-tier SCI international journals. Additionally, nine domestic and international patents were filed/registered, showcasing the innovation and practicality of the technology.
<(from left to right) Ph.D. candidates Song-I Cheon, Haidam Choi and Professors Minkyu Je >
The growing interest in wearable medical devices, which monitor physiological signals to analyze users’ health and predict potential diseases, has fueled advancements in various measurement technologies. One notable example is bio-impedance measurement, widely recognized for providing body composition data through “InBody.” With significant improvements in resolution, this technology can also capture plethysmography, enabling the measurement of heart rate information and further expanding its applications in health monitoring.
Recently, an international research team developed a new method to measure bio-impedance with five times greater resolution than existing technologies, using only two electrodes. This breakthrough, especially significant for the miniaturization of wearable devices, was the result of a collaboration between KAIST’s IMPACT Lab, led by Associate Professor Minkyu Je, and New York University Abu Dhabi (NYUAD), led by Associate Professor Sohmyung Ha.
Compared to the conventional four-electrode system, which is less suitable for compact devices, the two-electrode system offers significant advantages in miniaturization. However, previous two-electrode systems have faced challenges due to baseline impedance interference and increased noise proportional to the measured impedance.
Figure 1. (Left) An example of bio-impedance measurement using wearable devices; (Right) A conceptual diagram of the new impedance measurement circuit featuring baseline impedance cancellation and noise reduction functionality.
To overcome these limitations, the team developed an innovative semiconductor circuit design that effectively cancels out baseline impedance and associated noise. This design eliminates the need for a separate current generation circuit, reducing power consumption and improving measurement efficiency. The proposed system resolves noise issues caused by impedance phase and magnitude variations, achieving both high resolution and reliability.
“This bio-impedance measurement technology demonstrated up to five times better noise performance compared to conventional methods across various impedance models,” said Associate Professor Minkyu Je, a corresponding author of the work. “We believe this advancement will significantly contribute to the development of personalized health management and disease prediction technologies.”
The results of this work were published in the IEEE Journal of Solid-State Circuits, one of the most prestigious journals in the field of semiconductor integrated circuits and systems. The article, titled “A Bio-Impedance Readout IC With Complex-Domain Noise-Correlated Baseline Cancellation”, was featured in the journal in November 2024.
< Figure 2. Noise performance measurement results for various impedance models >
KAIST Ph.D. candidates Haidam Choi and Song-I Cheon were the co-first authors, while Professors Minkyu Je and Sohmyung Ha served as co-corresponding authors. The research was also presented at the International Solid-State Circuits Conference (ISSCC), the leading global conference in the field.
This work was funded by Korea’s Ministry of Science and ICT as part of projects focusing on continuous musculoskeletal monitoring and rehabilitation technologies, as well as the development of biosignal sensor-based exoskeleton devices and integrated systems.
Research Publication: IEEE Journal of Solid-State Circuits (2024), DOI:10.1109/JSSC.2024.3439865 Title: A Bio-Impedance Readout IC With Complex-Domain Noise-Correlated Baseline Cancellation
Professor Hye Won Chung has been selected as a Distinguished Lecturer by the IEEE Information Theory Society (ITSOC).
The IEEE Information Theory Society is the largest and most prestigious organization in the field of information theory. Each year, it selects 5 leading researchers as Distinguished Lecturers, offering them the opportunity to deliver invited lectures across approximately 50 chapters worldwide for two years. This program allows society members to learn about the latest research trends and outstanding achievements while engaging directly with the Distinguished Lecturers.
<An image representing graph matching research, which evaluates the similarity between two anonymized graphs, identifies one-to-one node correspondences, and expresses correlations between networks.>
Professor Hye Won Chung was recognized for her exceptional contributions to the fields of data science theory and data-efficient machine learning. She will serve as a Distinguished Lecturer from 2025 to 2026.
Furthermore, she has been invited to deliver a tutorial lecture titled “Graph Matching: Fundamental Limits and Efficient Algorithms” at the IEEE International Symposium on Information Theory (ISIT) 2024, the premier conference in information theory.
< Professor Sanghun Jeon, Winner of the 2024 Haedong Outstanding Paper Award >
Professor Sanghun Jeon from the Department of Electrical Engineering has been selected as the top recipient in the academic category of the 2024 Haedong Outstanding Paper Award, hosted by The Institute of Electronics and Information Engineers (IEIE) (President Chungyong Lee, Yonsei University) and sponsored by the Haedong Science Foundation.
The award ceremony took place on November 25 at 6 PM in the Grand Ballroom of the Convention Tower at the High1 Resort. The Haedong Awards by The Institute of Electronics and Information Engineers consist of three categories: Academic Award, Technology Award, and Young Engineer Award. The JSTS Academic Award specifically recognizes the best paper published in the JSTS journal over the past three years.
Professor Sanghun Jeon was recognized for his paper, “Ferroelectricity in Al2O3/Hf0.5Zr0.5O2 Bilayer Stack: Role of Dielectric Layer Thickness and Annealing Temperature,” which proposes dielectric stacking structures and design guidelines to enhance the thermal stability, performance, and reliability of hafnia ferroelectrics – a material gaining attention as a next-generation memory and storage device.
Once again, congratulations to Professor Sanghun Jeon for elevating the reputation of the Department of Electrical Engineering
< From left to right: Master’s candidate Woo-Hyun Hwang and undergraduate Jae-Jun Lee >
Master’s candidate Woo-Hyun Hwang and undergraduate student Jae-Jun Lee from Professor Seonghwan Cho’s lab achieved remarkable success by winning the Grand Prize at the ‘2024 Korean University Semiconductor Circuit Design Competition Award Ceremony’, held on December 11 at the Grand InterContinental Seoul Parnas Hotel.
The ‘Korean University Semiconductor Circuit Design Competition’, organized by the Institute of Semiconductor Engineers, aims to nurture IC circuit design skills among university students nationwide, discover creative ideas, and is supported by numerous semiconductor-related companies.
< From left to right: Jae-Jun Lee and Woo-Hyun Hwang at the ceremony>
At the awards ceremony on December 11, the Grand Prize-winning project titled ‘A Chip-Scale Operable Magnetic Resonance-Based Two-Coil Wireless Communication System’ was developed by Woo-Hyun Hwang and Jae-Jun Lee.
The project received high praise for its creativity, complexity, and level of completion. In particular, the innovative concept of implementing a “magnetic resonance-based two-coil system” at a chip scale, and the ability to achieve effective design with high completion in challenging wireless communication environments, were recognized and led to the Grand Prize award.
<(From left) Ph.D. Candidate Mintaek Oh, Professor Jinseok Choi>
Mintaek Oh, a Ph.D. Candidate in Professor Jinseok Choi’s lab at our department, achieved a Silver Award at the ‘2024 IEEE Student Paper Contest’ held by IEEE Seoul Section on December 7th.
The ‘IEEE Student Paper Contest’ is a prestigious competition organized by the IEEE Seoul Section that identifies outstanding research papers across all fields of electrical and electronic engineering and recognizes students’ research capabilities. Students who present outstanding papers at this year’s contest will be granted a privilege to participate in 2025 IEEE Region 10 Student Paper Contest.
<Mintaek Oh, recipient of the Silver Award at the ‘2024 IEEE Student Paper Contest’>
The award-winning paper titled “Multi-RIS-Aided Beamforming Design for MU-MIMO Systems with Imperfect CSIT” presents research on beamforming design considering practical constraints in wireless communication systems utilizing the RIS, which is emerging as a key technology for next-generation wireless communication systems. This research proposes an optimization approach for MU-MIMO system performance using multiple RISs under imperfect channel state information.
In this competition, amid competition spanning all fields of electrical engineering in Korea, the research excellence in wireless communications was recognized with the Silver Award.
A research team led by Professor Junmo Kim from the Department of Electrical Engineering at KAIST has developed an innovative AI technology that can envision and understand how images change, similar to how humans imagine transformations like rotation or recoloring. This breakthrough goes beyond simply analyzing images, enabling the AI to comprehend and express the processes involved in transforming visual data. The technology holds promise for diverse applications, including medical imaging, autonomous driving, and robotics, where precision and adaptability are essential.
AI That Imagine Changes Like Humans (Understands How Images Change, Like Humans)
The newly developed technology, Self-supervised Transformation Learning (STL), focuses on enabling AI to learn how images transform. STL operates without relying on human-provided labels; instead, it learns transformations by comparing original images with their transformed versions. It independently recognizes changes such as, “This has been rotated,” or, “The color has changed.” This process parallels the way humans observe, imagine, and interpret variations in visual data.
Illustration of the roles of the three representation learning approaches that constitute STL: (a) distinguishing images regardless of transformations, (b) aligning transformation representations for the same transformation applied to different images, and (c) ensuring that relationships between representations of transformed variants of the same image reflect the actual transformation. STL integrates all these roles for comprehensive learning
Overcoming the Limitations of Conventional Methods
Traditional AI systems often struggle with subtle transformations, focusing primarily on capturing large, overarching features while ignoring finer details. This limitation becomes a significant challenge in scenarios where precise understanding of intricate changes is crucial.
STL addresses this gap by learning to encode even the smallest transformations in an image into its feature space—a conceptual map representing the relationships between different data points. Rather than ignoring these changes, STL incorporates them into its feature representations, enabling more accurate and nuanced outcomes.
For example, STL excels at recognizing specific alterations, such as random cropping, brightness adjustments, and color modifications, achieving performance improvements of up to 42% over conventional methods. It is particularly adept at handling complex transformations that were previously difficult for AI to manage.
Examples of transformations often ignored by existing methods. These transformations, though subtle, may carry important differences.
Smarter AI for Broader Applications
What sets STL apart is its ability to not only understand visual content but also learn and represent transformations themselves. This capability allows STL to detect subtle changes in medical images, such as CT scans, and better interpret diverse conditions in autonomous driving. By incorporating transformations into its understanding, STL can deliver safer and more precise results across various applications.
Toward Human-Like Understanding
“STL represents a significant leap forward in AI technology, closely mirroring the way humans perceive and interpret changes in images,” said Professor Junmo Kim. “This approach has the potential to drive innovations in fields such as healthcare, robotics, and self-driving cars, where understanding transformations is critical.”
The research, conducted by Jaemyung Yu, a PhD candidate at KAIST as the first author, was presented at NeurIPS 2024, one of the world’s leading AI conferences, under the title Self-supervised Transformation Learning for Equivariant Representations. It was supported by the Ministry of Science and ICT through the Institute of Information and Communications Technology Planning and Evaluation (IITP) as part of the SW StarLab program (No. RS-2024-00439020, Development of Sustainable Real-time Multimodal Interactive Generative AI).
<(from left) Professor Chan-Hyun Youn, Ph.D. candidate Jinhyuk Jang>
Professor Chan-Hyun Youn’s research team has developed a novel approach to Machine Unlearning, a critical aspect of AI safety, by introducing the past weight prediction model InvWNN. This model aims to selectively remove the influence of problematic data from AI models trained on such data. Existing methods often require access to the entire training dataset or face challenges with performance degradation. To address these issues, the team proposed a new method that leverages weight history to predict past weights and iteratively applies this process to progressively eliminate the influence of the data.
<Figure 1: Machine Unlearning Process of the Proposed InvWNN and Unlearning Trajectory>
This method gradually removes the problematic data’s influence through iterative fine-tuning on the problematic data and the past weight prediction. Notably, this approach operates effectively without access to the remaining data and can be applied to various datasets and architectures. Compared to existing methods, the proposed technique excels at accurately removing unnecessary knowledge from training data while minimizing side effects. Furthermore, it has been validated that this method can be directly applied to a variety of tasks without requiring additional procedures.
The research team demonstrated the high performance of their method across various benchmarks, showcasing its potential to significantly expand the practical applicability of machine unlearning technology. These findings will be presented at the 39th Annual AAAI Conference on Artificial Intelligence (AAAI 2025), one of the premier international conferences in the field of artificial intelligence, to be held in the United States in February next year, with the title “Learning to Rewind via Iterative Prediction of Past Weights for Unlearning.”
