-
Designable van der Waals Crystal Realizes Artificial Neuronal Cell Mimicking with Light
A research team led by Professor Taesung Kim of the School of Mechanical Engineering developed an optoelectronic synaptic device that mimics the functions of human neurons and synapses at the device scale. The researchers designed a designable van der Waals (vdW) crystal through a single-step sulfurization process using mixed plasma. The developed device operates under optical stimuli, offering a structural solution to configure semiconductor materials for brain-inspired computing. Rapid advancements in artificial intelligence and hyper-connectivity require neuromorphic vision systems capable of sensing and processing vast amounts of visual data in real time. Optoelectronic synapses, which exhibit conductance variations in response to light signals, serve as core components of these systems. Layered vdW materials attracted significant attention as promising candidates due to their excellent optical properties and atomic-scale thickness. However, conventional vdW materials faced technical challenges, including the difficulty of precisely controlling grain boundaries and intercalation, polymer residue accumulation, mechanical warpage at interfaces, and poor large-area crystalline uniformity. To overcome these limitations, the research team focused on the structural similarity between light-sensitive ion channels in biological membranes and layered vdW lattices. The researchers applied an argon and hydrogen sulfide (Ar + H₂S) plasma sulfurization process to bulk van der Waals rhenium selenide (ReSe₂). This single-step process transformed the upper portion of the material into a nano-crystalline ReSe₂ layer composed of nano-sized grains, while preserving the underlying bulk single-crystalline ReSe₂ layer without damaging the interlayer interfaces. These two integrated layers structurally correspond to the light-sensitive ion channels of a neuronal cell membrane and the intracellular environment, respectively, and were fabricated without additional deposition or patterning steps. The research team utilized scanning probe microscopy (SPM) to resolve the pathways of S²⁻ (sulfur) ionic migration. The grain boundaries in the nano-crystalline ReSe₂ layer confined the sulfur ionic transport at the atomic scale, enabling deterministic control over synaptic weight updates, similar to the gating mechanism of biological ion channels. The device demonstrated key synaptic functionalities, including multi-level conductance modulation, long-term potentiation/depression (LTP/LTD), paired-pulse facilitation (PPF), and a tunable short-term to long-term memory (STM-LTM) transition. The nano-crystalline ReSe₂ device exhibited a 34.7% increase in retention efficiency during learning-forgetting-relearning cycles compared to bulk ReSe₂. In system-level evaluations, the device successfully performed edge detection on natural images and achieved a 96.24% classification accuracy on the CIFAR-10 image recognition task. This development offers a materials platform for next-generation neuromorphic semiconductors and AI hardware. "This study demonstrates a single-step method to design the structure of van der Waals crystals for optoelectronic synaptic devices that learn and store information using light," said Professor Taesung Kim, the corresponding author of the study. "By structurally resolving the random nature of ionic migration and interfacial issues inherent in conventional devices, this architecture can be applied to research on next-generation neuromorphic semiconductors and AI hardware." This research received financial support from the National Research Foundation of Korea (NRF) Leader Research Program, the Institute for Basic Science (IBS), and the Semiconductor-Track Graduate School Program funded by the Ministry of Trade, Industry and Energy (MOTIE). The study was conducted as a collaborative effort among researchers from Sungkyunkwan University (SKKU), the Center for Quantum Nanoscience at IBS, and the Korea Institute of Machinery and Materials (KIMM). The findings were published online in the international journal Advanced Materials (Impact Factor: 26.8, top 1% in JCR) on June 3, 2026. ▲ Structure and operational mechanism of the optoelectronic synaptic device based on the designable van der Waals crystal operating with light published in advanced materials SKKU RESEARCH STORY Designable van der Waals Crystal for Artificial Neuronal Cell Mimicking Access Publication (DOI) TK Taesung Kim PURE Profile →
- No. 380
- 2026-06-15
- 351
-
Ultralong-Life Aqueous Batteries Enabled by Nanostructured Electrolyte Additives
A novel electrolyte technology that overcomes the persistent limitations of aqueous batteries has been developed by a Korean research team. The research team led by Professor Hoseok Park of the Department of Chemical Engineering announced that they have succeeded in dramatically improving the cycle life and capacity of aqueous batteries simply by adding a small amount of a special material to the electrolyte. Aqueous batteries have long been considered a promising alternative for energy storage systems due to their lower raw material costs compared to lithium-ion batteries, as well as their superior safety and environmental friendliness stemming from the use of water-based electrolytes. However, the uneven deposition of zinc on electrode surfaces during repeated charge-discharge cycles, along with parasitic side reactions between zinc metal and water in the electrolyte, have accelerated battery degradation and posed a major obstacle to commercialization. To address these issues, the research team turned their attention to zwitterions — molecules that carry both a positive and a negative charge simultaneously within a single molecular structure. Although electrically neutral overall, zwitterions possess the unique ability to precisely regulate interactions with surrounding ions. The zwitterionic additive developed by the team (C10) spontaneously self-assembles into nanostructures approximately 3.77 nm in diameter when introduced in small quantities into the electrolyte. These nanostructures serve two critical functions. First, they guide zinc ions to deposit uniformly and stably onto the electrode surface, suppressing uneven zinc plating. Second, they coat the zinc metal surface with a thin, uniform protective layer that effectively blocks parasitic side reactions with water and prevents corrosion. Aqueous batteries incorporating the developed electrolyte achieved an ultralong-term cycling stability of over 2,800 hours of stable operation. Under high-capacity conditions, the batteries recorded an areal capacity of 8.10 mAh cm⁻², representing world-leading performance among aqueous batteries reported to date. The simultaneous improvement of both cycle life and capacity — two critical performance metrics — stands out as the defining achievement of this study. Professor Hoseok Park remarked, "We have demonstrated that the performance of aqueous batteries can be substantially enhanced through a simple approach of adding a small amount of material to the electrolyte, without the need for expensive materials or complex fabrication processes." He added, "Beyond renewable energy storage, this technology holds potential for application in large-scale energy storage systems (ESS) for AI infrastructure and data centers, which are experiencing explosive growth." This research was supported by the Leader Research Program and the Future-leading Pioneer Research Program funded by the Ministry of Science and ICT and the National Research Foundation of Korea. The findings were published on January 4th in Nano-Micro Letters (IF 36.3, top 1% in nanotechnology). Schematic illustration of the role of C10 in zinc-ion batteries: formation of localized high-concentration electrolytes through C10 aggregation and induction of uniform zinc electrodeposition via electric double layer self-assembly, along with suppression of parasitic side reactions. Published in nano-micro letters SKKU RESEARCH STORY Self-Assembled Ordered Nanostructure of Zwitterionic Co-Solutes Induces Localized High-Concentration Electrolytes for Ultrastable and Efficient Zinc Metal Anodes Access Publication (DOI) HP Prof. Hoseok Park PURE Profile →
- No. 379
- 2026-06-10
- 688
-
The Rise of Meme Advertising: Why Are Luxury Brands More Shareable?
Luxury brands have increasingly embraced internet memes in their social media advertising campaigns. For example, Gucci’s #TFWGucci campaign demonstrated how luxury brands can incorporate memes into their digital communication strategies to capture consumer attention. Memes—digital content that spreads rapidly through sharing, imitation, and adaptation across online communities—have become a popular way for brands to engage consumers. Branded memes, in particular, leverage humor, relatability, and virality to communicate brand messages in a more engaging and shareable format. However, the use of memes by luxury brands may seem somewhat unexpected, given that these brands have traditionally emphasized exclusivity, prestige, and sophistication. Although many companies actively employ meme advertising to connect with consumers, it remains unclear whether this communication strategy enhances consumer responses or weakens a brand’s premium image. Professor Tae Hyun Baek’s research team in the Department of Media and Communication at Sungkyunkwan University conducted an international collaborative research project with Professor Jooyoung Park from Peking University HSBC Business School in China. Across four experiments, they examined how luxury-branded meme advertising influences consumer responses. The findings were published in the International Journal of Advertising, a leading SSCI journal ranked among the top 1.5% of Communication journals according to the 2024 Journal Citation Reports (JCR). The results showed that consumers perceived luxury-branded memes as funnier than non-meme advertisements, with perceived unexpectedness mediating this effect (Study 1). Luxury-branded memes also enhanced consumers’ social media sharing intentions. Consumers viewed the combination of luxury brands and memes as unexpected, and the funnier they found the content, the more likely they were to share it with others (Study 2). The same pattern emerged in a field experiment using Facebook A/B testing. Luxury-branded meme ads generated more clicks and user engagement than non-meme ads (Study 3). However, meme advertising was not equally effective for all brands. While meme ads increased consumers’ sharing intentions for luxury brands such as Prada, non-meme ads generated stronger sharing intentions for non-luxury brands such as Zara (Study 4). According to Professor Baek, “Luxury-branded memes appear to be most effective when consumers perceive the ad content as both unexpected and funny. Our findings suggest that luxury brands should carefully align meme advertising strategies with their brand positioning and communication objectives.” Published in international journal of advertising SKKU RESEARCH STORY Meme advertising for luxury brands: Effects on perceived funniness and sharing intention Access Publication(DOI) TB Tae Hyun Baek PURE Profile →
- No. 378
- 2026-06-05
- 707
-
Development of an Ultra-Stretchable Anti-Freezing hydrogel electrolyte based on liquid metal
The research group led by Prof. Sungjune Park from the Department of Chemical Engineering has developed an ultra-stretchable, anti-freezing hydrogel electrolyte using liquid metal particles. The material can stretch up to nine times its original length while maintaining stable electrochemical performance, even at −20 °C. This work provides a promising platform for energy storage devices that must operate reliably under extreme environmental conditions. With the rapid growth of wearable electronics, there is increasing demand for energy storage systems that combine mechanical flexibility with environmental stability. However, conventional hydrogel electrolytes typically suffer from low mechanical strength and freezing at low temperatures, leading to significant performance degradation. The research group used liquid metal particles as an initiator for polymerization. Under ultrasonication, the bulk liquid metal was broken into fine particles, which then initiated the polymerization of acrylamide and acrylic acid to form the hydrogel. This process eliminates the need for external stimuli such as heat or ultraviolet irradiation, simplifying fabrication and improving scalability. The group added stearyl methacrylate (SMA), a hydrophobic material that does not mix well with water, to create physical crosslinking between polymer chains. These physical crosslinks act as reversible connections within the network. When an external force is applied, the bonds can break to dissipate energy and then easily reform once the stress is released, thereby imparting exceptional stretchability and mechanical robustness to the material. As a result, the elongation at break (defined as the maximum stretch before the material fails) reached up to 900% of its original length. After soaking the hydrogel in a lithium chloride (LiCl) solution, it exhibited anti-freezing properties by suppressing hydrogen bonding between water molecules. It maintains both ionic conductivity and mechanical flexibility even at −20 °C, where conventional hydrogel systems typically fail. Energy storage devices fabricated with this electrolyte retained 98% of their performance after 45,000 charge-discharge cycles. The research group noted, “For practical applications, it is essential to ensure long-term stability and reproducibility in large-area manufacturing processes.” Prof. Park stated, “This work introduces a new design strategy for hydrogel electrolytes based on liquid metal and provides a viable platform for next-generation wearable electronics and flexible energy storage systems operating under extreme conditions.” The research results were published on March 13 in Nano-Micro Letters. ▲Schematic illustration of the fabrication process and device structure of the liquid metal-based hydrogel electrolyte Published in Nano-Micro Letters SKKU RESEARCH STORY Ultra-Stretchable Anti-Freezing Hydrogel Electrolytes Cross-Linked by Liquid Metal Particle Initiators Toward Soft Energy Storage Devices Access Publication (DOI) SP Sungjune Park PURE Profile →
- No. 377
- 2026-06-02
- 1097
-
Development of Bayesian Inference for Hidden Dependence Structures in Multi-Group High-Dimensional Data
The research team of Professor Kyoungjae Lee of the Department of Statistics at Sungkyunkwan University, through joint research with Professor Won Chang of Seoul National University and Professor Xuan Cao of the University of Cincinnati, developed Bayesian inference for the hidden dependence structures of multi-group high-dimensional data. A Dependence Map in High-Dimensional Data In today’s scientific and industrial fields, high-dimensional data in which numerous variables are observed simultaneously, such as genomic, climate, financial, and sensor data, are rapidly increasing. In such data, an important problem is to learn the dependent structures connecting the variables and to identify a “dependence map” that reveals hidden information in massive datasets. For example, in climate data, temperatures in nearby regions may be related to one another, and in genomic data, genes located in adjacent positions may act together. If such dependence can be incorporated into inference, more efficient inference is possible than analyzing each variable separately. Development of the j-LANCE Method for Joint Inference of Dependence Across Multiple Groups The j-LANCE (joint LocAl depeNdence CholEsky) method proposed in this study focuses on the fact that, in real data such as genomic and climate data, variables have a natural ordering and are mainly related to nearby neighboring variables. Based on this idea, the method estimates the extent to which each variable is connected to neighboring variables and is designed to learn similar structures across multiple groups while allowing group-specific differences. In many existing methods, data from multiple groups are either analyzed separately or simplified by assuming that all groups have the same structure. In contrast, this study uses a Markov random field prior so that similarities and differences across groups can be flexibly learned from the data. Simultaneously Achieving Theoretical Accuracy and Fast Computation An important achievement of this study is that it simultaneously attains theoretical accuracy and computational efficiency even in high-dimensional settings. This study theoretically proved that j-LANCE can accurately estimate the dependence structures of multiple groups, and also showed that the rate at which the estimates approach the true values is nearly minimax-optimal. In addition, the methodology was designed to enable Bayesian inference without using MCMC, a complex iterative computation procedure, thereby securing the advantage that fast analysis is possible even for high-dimensional data. Practical Applicability Confirmed Through Climate Data Analysis In this study, ERA5 data were used to analyze temperatures at 30 locations in the Pacific Northwest region of the United States from 2019 to 2021, and the dependence structure of temperatures across regions was estimated based on a spatial ordering that reflects wind flow. As a result, j-LANCE was found to capture similar dependence patterns across years while also detecting distinctive dependence structures that appeared in a specific year. This confirmed the practical applicability of j-LANCE to real data, and the method is expected to be applicable in a wide range of fields that require simultaneous analysis of complex data from multiple groups, including climate, genomics, finance, and sensor time series. *This research achievement was published in Bayesian Analysis, an international journal in the field of statistics. Published in bayesian analysis SKKU RESEARCH STORY The Joint Local Dependence Cholesky Prior for Bandwidth Selection Across Multiple Groups Access Publication (DOI) KL Kyoungjae Lee Professor Profile → Figure 1. Temperature heatmap of ERA5 climate data Figure 2. Cholesky factors estimated from ERA5 climate data. Each column corresponds, from right to left, to 2019, 2020, and 2021, and each row shows, from top to bottom, the results of j-LANCE, year-specific independent LANCE, the penalized likelihood-based inference method, and the group graphical lasso method
- No. 376
- 2026-05-26
- 999
-
SKKU Research Team Develops Experimentally Validated AI Model to Predict the Virulence of Tomato Yellow Leaf Curl Virus
A CBBL research team led by Professor Balachandran Manavalan from the Department of Integrative Biotechnology at Sungkyunkwan University has developed DeepTYLCV, an accurate and interpretable artificial intelligence model for predicting the virulence of Tomato Yellow Leaf Curl Virus (TYLCV). The study was conducted with co-first authors Dr. Nattanong Bupi, Hariharan Sangaraju, and Duong Thanh Tran, was published in the leading plant science journal Plant Communications (Impact Factor: 11.6; JCR: 6/273; Top 2.2% in the Plant Sciences category). TYLCV is one of the most destructive viral pathogens affecting tomato production worldwide. Severe TYLCV strains can cause leaf curling, yellowing, stunted growth, and major yield losses. In recent years, highly virulent strains have continued to spread across regions and, in some cases, overcome genetic resistance in tomato cultivars. These challenges highlight the urgent need for accurate, early, scalable, and sequence-based disease surveillance. Prof. Manavalan’s team has been working extensively at the interface of biology and artificial intelligence, developing AI-based solutions for peptide therapeutics, prediction of RNA/DNA modifications, protein function analysis, toxicity prediction, plant science, and biomedical applications. In 2023, the team developed IML-TYLCV, the first genome-based TYLCV severity prediction tool, which was published in the high-impact journal Research (IF: 10.9). However, IML-TYLCV was mainly trained on Korean isolates, limiting its applicability to globally diverse TYLCV strains. This challenge motivated the development of DeepTYLCV, a more robust AI framework for predicting TYLCV virulence across global viral isolates. Unlike conventional field diagnosis or image-based AI models, which depend on visible symptoms and can be influenced by environmental conditions, DeepTYLCV uses viral genome-derived sequence information. This enables the model to identify mild and severe strains before symptom-based confirmation and provides a scalable strategy for monitoring emerging viral variants. DeepTYLCV integrates protein language model embeddings with a hybrid architecture that combines a Transformer encoder and a multi-scale convolutional neural network, enabling the model to capture both global sequence patterns and local virulence-associated motifs. By combining deep sequence representations with optimized conventional feature descriptors, DeepTYLCV achieved superior predictive performance compared with the previous IML-TYLCV model. A key strength of this study is its experimental validation. The research team performed blind predictions for 15 TYLCV isolates, including both international reference isolates and Korean field isolates. These predictions were validated using tomato plant infection assays, symptom severity scoring, and viral accumulation analysis. Remarkably, DeepTYLCV achieved 100% concordance between predicted and experimentally observed virulence classes, demonstrating its practical value for identifying emerging severe TYLCV variants. This work provides a powerful example of how AI, viral genomics, and experimental plant pathology can be integrated to support precision agriculture and plant disease management. DeepTYLCV may serve as a valuable tool for early viral surveillance, resistance breeding programs, and rapid assessment of newly emerging TYLCV strains. This research was supported by the National Research Foundation of Korea (NRF), funded by the Ministry of Science and ICT, Republic of Korea (Grant No. RS-2024-00344752). Additional support was provided by the BK21 FOUR Project of the Department of Integrative Biotechnology, Sungkyunkwan University (SKKU), Republic of Korea. Published in plant communications SKKU RESEARCH STORY DeepTYLCV: An interpretable and experimentally validated AI model for predicting virulence of different tomato yellow leaf curl virus strains Access Publication (DOI) BM BALACHANDRAN MANAVALAN PURE Profile→ Figure 1. Overview of the DeepTYLCV framework. The framework involves six key stages: (A) Collection and pre-processing of global TYLCV genomes into open reading frames. (B) Projection and stacking of PLM/NLP embeddings to capture sequence context. (C) A hybrid Transformer encoder and Multi-scale CNN module for learning global and local virulence patterns. (D) Selection of optimal conventional descriptor features. (E) A multi-layer perceptron (MLP) classifier for severity prediction. (F) Deployment of a user-friendly web server. Figure 2. Experimental validation of DeepTYLCV predictions through symptom development and viral quantification in tomato plants. Plants were agro-inoculated with 15 TYLCV infectious clones. (A) Prediction probabilities from the current method, DeepTYLCV, and the previous method, IML-TYLCV. (B) Viral DNA accumulation at 21 days. (C) Symptom severity was monitored for 21 days. (D) Visual symptoms of infected plants at 21 days. (E) PCR detection confirming viral infection.
- No. 375
- 2026-05-15
- 2255
-
Professor Junghee Bae’s Research Team Identifies Performance Differences in Social Enterprises by Legal Form
The research team led by Professor Junghee Bae from the Department of Social Welfare at Sungkyunkwan University analyzed the full population dataset of certified social enterprises in South Korea to compare the social and economic performance of Work Integration Social Enterprises (WISEs) across different legal forms—nonprofit, for-profit, and cooperative organizations. Work Integration Social Enterprises are organizations that pursue both social and economic goals by providing employment opportunities to vulnerable populations who are marginalized in the labor market, while generating revenue through business activities. These enterprises play a crucial role in promoting social integration and economic self-sufficiency among disadvantaged groups. In South Korea, under the Social Enterprise Promotion Act enacted in 2007, organizations must adopt legally recognized forms—such as corporations, social welfare foundations, nonprofit organizations, or cooperatives—in order to receive official certification as social enterprises. The findings reveal that even among social enterprises with the same objective of creating jobs for vulnerable populations, performance varies significantly depending on legal form. In particular, nonprofit social enterprises were found to employ a larger number of vulnerable individuals and to have a higher proportion of such employees in their workforce, indicating stronger social performance compared to for-profit and cooperative counterparts. Moreover, nonprofit social enterprises demonstrated relatively higher net income by leveraging diverse revenue sources, including government subsidies, private donations, and public-sector market sales. In contrast, for-profit and cooperative social enterprises showed relatively higher working hours and wage levels for vulnerable employees. However, these organizations tended to rely more heavily on private market revenues, which was associated with comparatively lower overall financial performance. This study highlights that the choice of legal form plays a critical role in shaping both social value creation and financial sustainability in social enterprises, providing empirical support for institutional theory, which emphasizes the influence of institutional environments on organizational performance. In particular, amid the recent rapid growth of corporation-type social enterprises in South Korea, the findings suggest the need for balanced ecosystem development and stronger policy support for nonprofit social enterprises, especially in light of their effectiveness in achieving the core mission of employing vulnerable populations. The study was published in the leading international journal in the nonprofit field, Nonprofit and Voluntary Sector Quarterly, Volume 55, Issue 2. Published in nonprofit and voluntary sector quarterly SKKU RESEARCH STORY Work integration social enterprises with different legal forms: Performance comparison between nonprofit, for-profit, and cooperative organizations Access Publication (DOI) JB Junghee Bae Profile →
- No. 374
- 2026-05-15
- 840
-
From Common Natural Sweetener to High-Performance Energy Material
Professor Kyungwho Choi’s team (co-first authors: Thien Trung Luu and Bui Minh Quang) of the School of Mechanical Engineering at Sungkyunkwan University, in collaboration with Professor Jinsoo Kim’s team in the Department of Chemical Engineering at Kyung Hee University, proposed a strategy that simultaneously overcomes the limitations of conventional hydrogel-based triboelectric nanogenerators (TENGs) — namely low output performance, poor mechanical strength, and insufficient transparency — by utilizing biomimetic stevia. By incorporating stevia into polyvinyl alcohol (PVA), the abundant hydroxyl groups (-OH) simultaneously reinforced the hydrogen bond-based crosslinking structure and crystalline domains, dramatically improving both mechanical strength and ionic conductivity. As a result, the stevia-PVA hydrogel TENG (S-TENG) demonstrated approximately 2–5 times greater mechanical strength and 3–8 times higher electrical output compared to conventional TENGs based on 2D materials, biomaterials, and transparent materials, while maintaining over 70% visible light transmittance. The tensile strength exceeded 25 MPa (in the hydrated state) with an elongation at break surpassing 510%. Furthermore, the research team demonstrated that the S-TENG maintained stable output (~800 V) through 16,000 contact-separation cycles, and confirmed no degradation in electrical output after 30 days of storage at room temperature. The stevia hydrogel can also be recycled via a water-assisted dissolution and re-gelation process, retaining a high output voltage of approximately 600 V after recycling, thus demonstrating its potential as an eco-friendly material. In addition, the research team attached the S-TENG to various body parts — including the wrist, elbow, knee, finger, and throat — and utilized it as a self-powered sensor for detecting diverse human body motions. The rise time in response to finger bending was as fast as 13 ms, and among eleven machine learning models evaluated for motion classification, the XGBoost algorithm achieved the highest classification accuracy of 95.29%. Professor Kyungwho Choi, the corresponding author, stated: "It is highly significant that we successfully developed a hydrogel electrode derived from biomass-based stevia that simultaneously improves transparency, mechanical performance, and electrical output while also securing recyclability. We plan to continue research on applying this technology to a wide range of fields, including IoT-based wearable devices, rehabilitation monitoring, and intelligent human-machine interfaces." This research was supported by the 4th BK21 Future HRD Education and Research Center for Human-Centered Convergence Mechanical Solution and by the Korea government (MSIT). The results were published in Advanced Materials(IF 26.8, within the top 3% of JCR) in April 2026. In addition, this paper was selected for the inside front cover of Advanced Materials. ▲Schematic diagram of the structure and motion recognition system of a stevia-enhanced PVA hydrogel-based wearable sensor ▲ Selected as the Inside Front Cover article in the journal Advanced Materials Published in Advanced Materials High-Performance Transparent, Deformable, and Recoverable Biomimetic Stevia–PVA Hydrogel Triboelectric Nanogenerator with Machine Learning-Assisted Motion Recognitions Access Publication (DOI) KC Kyungwho Choi PURE Profile →
- No. 373
- 2026-05-11
- 1590
-
Does Personalized Virtual Try-On Turn Imagination into Reality?
When shopping for clothing online, how confident can we be without actually trying the product on? In digital shopping environments, consumers often experience uncertainty—such as “Will this really suit me?”—due to the inability to physically interact with products. To address this limitation, virtual try-on technology has emerged. More recently, beyond basic virtual fitting, personalized virtual try-on technologies that reflect individual body shapes and styles have been rapidly advancing. Professor Seeun Kim of Sungkyunkwan University, in collaboration with a research team from Oklahoma State University, conducted an empirical study examining the impact of personalized virtual try-on on consumer decision-making. The research focused particularly on how easily consumers can imagine products and how this process contributes to psychological confidence in purchase decisions. The findings revealed that personalized virtual try-on significantly enhances product imagination. By viewing virtual images that closely resemble their own bodies, consumers are able to vividly imagine themselves wearing the product. This imagination extends beyond a simple cognitive process and directly influences decision-making. The more vividly consumers can imagine a product, the more comfortable and confident they feel about their choices. This suggests that virtual try-on reduces uncertainty—such as “Will this product suit me?”—and facilitates more stable decision-making. Interestingly, these effects were not uniform across all consumers. The study identified spatial processing perception—an individual cognitive trait—as a key moderating factor. The effects of virtual try-on were stronger among consumers with lower spatial processing ability. This can be explained by the fact that individuals who have difficulty mentally visualizing products rely more heavily on the visual information provided by virtual try-on. In contrast, consumers with higher spatial processing ability can already imagine products effectively, making the additional benefits of virtual try-on relatively limited. In other words, personalized virtual try-on is not merely a “better technology,” but rather a technology that is more beneficial for certain consumers. This study is meaningful in that it uncovers the underlying mechanism through which virtual try-on goes beyond visual experience to influence consumers’ psychological decision-making processes. By identifying how product imagination translates into decision comfort, the research provides important implications for designing consumer experiences in online shopping environments. Looking ahead, fashion and e-commerce companies should move beyond simply adopting new technologies and instead develop personalized strategies that align with consumers’ cognitive characteristics to deliver more effective shopping experiences. The study has been published in the internationally recognized SSCI journal Journal of Research in Interactive Marketing. Published in Journal of Research in Interactive Marketing SKKU RESEARCH STORY Unveiling product imagination and decision comfort through personalized virtual try-on: the moderating role of spatial processing perception Available Access Publication (DOI) SK Seeun Kim SKKU Professor →
- No. 372
- 2026-05-07
- 1334
-
Revolutionizing Clinical Trials with Machine Learning
Professor Yeonhee Park of the Department of Statistics at Sungkyunkwan University has developed a novel statistical framework — MARGO (Machine Learning-Assisted Adaptive Randomization for Group Sequential Trials Based on Overlap Weights) — that makes machine learning practically applicable in clinical trial design. This work provides the first rigorous solution to the fundamental statistical challenges that arise when integrating ML/AI-driven decision-making into the scientifically demanding environment of clinical trials. The Promise and the Barrier: Why ML/AI Alone Is Not Enough Machine learning and artificial intelligence have garnered widespread attention as transformative tools for personalized treatment assignment in clinical trials. In particular, adaptive randomization — which dynamically adjusts treatment allocation based on accumulating trial data — is a promising approach for improving patient outcomes by steering more participants toward more effective treatments. However, applying this approach in practice can introduce a critical statistical problem. When patient characteristics (e.g., biomarkers) are used to guide treatment assignment, systematic imbalances can emerge between treatment groups. This covariate imbalance leads to biased treatment effect estimates and an inflated type I error rate, risking false conclusions. The problem is further compounded in group sequential designs, which include planned interim analyses for early stopping decisions. Machine Learning Meets Causal Inference: A Two-in-One Solution To address this fundamental challenge, MARGO integrates machine learning-based predictive models with overlap weights (OW), a propensity score–based approach widely used in causal inference to adjust for covariate imbalance. MARGO uses patient covariate information to predict the probability of treatment success via machine learning, then uses these predictions to preferentially assign patients to the more effective treatment. Simultaneously, OW corrects covariate imbalance across treatment groups, effectively controlling the bias and type I error inflation induced by adaptive randomization. The framework was evaluated using four machine learning algorithms: Support Vector Machine (SVM), K-Nearest Neighbors (KNN), Random Forest (RF), and Multi-Layer Perceptron (MLP). Rigorously Validated Performance Through extensive simulation studies, MARGO demonstrated superior performance over conventional fixed randomization and existing adaptive randomization methods across three key dimensions. First, MARGO allocated a greater proportion of patients to the more effective treatment. Second, it maintained the overall type I error rate below the target threshold of 0.05 — even in scenarios where conventional methods inflated the error rate to as high as 0.08–0.18. Third, it preserved high statistical power under alternative scenarios while reducing the number of treatment failures. Together, these results demonstrate that MARGO can simultaneously improve the ethical standards and scientific integrity of clinical trials. Beyond "Using AI" — Toward "Trusting AI in Clinical Trials" The most important contribution of this research goes beyond simply applying machine learning to clinical trials — it rigorously resolves the fundamental statistical problems that emerge in that process. MARGO is designed to accommodate a wide range of AI models and holds broad potential for extension to precision medicine and data-driven decision-making across diverse fields. This study was published in Statistics in Medicine in 2025. Published in Statistics in Medicine (2025) SKKU RESEARCH STORY MARGO: Machine Learning-Assisted Adaptive Randomization for Group Sequential Trials Based on Overlap Weights Access Publication (DOI) YP Yeonhee Park SKKU Professor → Figure 1. Design Framework for Adaptive Randomization with Interim Analyses Figure 2. Simulation results: Controlled Type I Error Rates Under the Nominal Level
- No. 371
- 2026-05-07
- 898
-
SKKU Revolutionizes Battery Manufacturing with Density Dry Electrode Technology; Targets Foundry Commercialization
A research team led by Professor Young-Jun Kim at the Sungkyunkwan Advanced Institute of Nano Technology (SAINT) of SKKU has announced a breakthrough in "Dry Electrode" technology-a next-generation manufacturing process for batteries. The team has secured original technology for electrodes with world-class energy density by developing materials optimized for solvent-free electrode processing, a feat expected to shift the paradigm of battery production. Dry electrode technology eliminates the use of toxic liquid solvents in the production of lithium-ion and all-solid-state batteries. By directly compacting solid raw materials into electrode films, the process removes the need for energy-intensive drying stages, making it both eco-friendly and highly cost-efficient. This field is currently a focal point for global industry leaders, including Tesla, as companies race to secure dominance in the future battery market. To overcome the chronic challenges of dry processing-specifically the difficulty of uniform mixing and large-scale production-Professor Kim’s team developed a "One-body" material that integrates active materials (for energy storage) and conductive agents (for electron conduction) into a single architecture. This innovation enables the mass production of high-quality, high-loading electrodes. The technical reliability and performance of this method were rigorously validated through collaborative simulations with Professor Yong-Min Lee’s team at Yonsei University. "Dry electrode technology is more than just an eco-friendly process; it is the ultimate solution to dramatically enhancing battery performance, quality, and safety," said Professor Young-Jun Kim. "The specialized materials and production techniques we've developed will serve as a critical stepping stone to significantly reducing manufacturing costs while ensuring global competitiveness in battery performance.“ Beyond academic achievement, the team is aggressively pursuing commercialization. Through Corenergy Solution, a laboratory-backed startup, the team plans to launch a "Battery Electrode Foundry" business specializing in dry electrode manufacturing. In collaboration with other SKKU faculty members with extensive industry experience at South Korean battery giants such as Samsung SDI and LG Energy Solution, the venture aims to advance dry electrode design and cell manufacturing technologies to strengthen the domestic battery ecosystem. This research was supported by the Nano-Material Technology Development Program through the National Research Foundation of Korea (NRF). The team's findings on dry cathode technology were published in Joule (IF 35.4), a premier global journal in the energy field, while their research on anodes was featured in the online edition of Carbon Energy (IF 24.2), gaining worldwide academic recognition. SKKU RESEARCH STORY A continuous carbon nanotube sheath enables ultrahigh energy density and fast charging in dry-processed thick electrodes Joule (DOI) Dry-Processed Graphite Electrodes Enabling Ultra-High Areal Capacity and Stable Fast-Charging Performance Carbon Energy (DOI) YK Young Jun Kim PURE Profile → ▲ Overview of Dry Electrode Manufacturing Process and Breakthrough Material Technologies for Battery Innovation
- No. 370
- 2026-04-28
- 1403
-
Prof. Eun Ryung Lee Receives the 2025 Korea Statistical Researcher of the Year for Her Annals of Statistics Paper
Professor Eun Ryung Lee of the Department of Statistics at Sungkyunkwan University received the 2025 2nd Korea Statistical Researcher of the Year in recognition of her paper “Efficient Functional Lasso Kernel Smoothing for High-Dimensional Additive Regression”, which was published in Annals of Statistics in August 2024. The award ceremony was held on August 28, 2025, at the 14th National Statistics Development Forum in Seoul. This award, presented by Statistics Korea, honors outstanding researchers who have made significant contributions to the development of statistics. The paper addresses a fundamental challenge in modern data analysis: how to identify truly important variables and accurately estimate their nonlinear effects when the number of variables is much larger than the sample size. In ultra-high-dimensional settings, it is difficult to achieve variable selection, flexible modeling, computational feasibility, and statistical inference at the same time. This study provides a new solution to that problem. Professor Lee and her collaborators developed a new kernel-based methodology that combines functional Lasso with smooth backfitting. The proposed method can automatically select important variables while flexibly estimating their effects through nonlinear functions, and it is supported by both computationally efficient algorithms and rigorous theoretical analysis. In addition, the study introduces a debiased inference procedure, making it possible not only to improve prediction accuracy but also to construct confidence intervals and conduct hypothesis testing. The proposed method was further applied to large-scale gene expression and anticancer drug response data from cancer cell lines, where it showed strong empirical performance and successfully identified biologically meaningful genes associated with drug response. The study is expected to have broad impact in bioinformatics, precision medicine, finance, environmental science, and other fields where high-dimensional data are increasingly common. This award recognizes both the originality and the practical importance of Professor Lee’s contribution to modern statistical methodology. Published in Annals of Statistics SKKU RESEARCH STORY Efficient functional Lasso kernel smoothing for high-dimensional additive regression Annals of Publication (DOI) EL Eun Ryung Lee Profile→ ▲ A three-stage graphical abstract — "input → method → output" summarizing the core idea of fLasso-SBF Left (Input) A grid of scatter plots for many candidate covariates, with the total number allowed to be much larger than the sample size. Most covariates (grey) carry essentially no information about the response and behave like pure noise, while only the three highlighted ones (red, green, blue) are truly active with genuine nonlinear effects. This depicts the high-dimensional sparse additive regression setting that motivates the paper. Middle (Method) The proposed fLasso-SBF method minimizes an objective that combines kernel-based smooth backfitting with a functional Lasso penalty. Its solution is obtained by iteratively applying a simple "soft-threshold + projection" update, which adds only a single thresholding step to the standard smooth backfitting algorithm and keeps both the implementation and the theoretical analysis clean. Right (Output) Overlay of the estimated component functions produced by fLasso-SBF. Only the three genuinely active components are recovered as smooth curves, while the estimates for the remaining inactive covariates are automatically shrunk to zero. Variable selection and nonparametric function estimation are thus carried out simultaneously in a single procedure, and the accompanying debiased version further enables pointwise confidence intervals and hypothesis testing.
- No. 369
- 2026-04-28
- 1067
