사업단소식 게시판목록 | Enterprise Organization of National Program of Excellence in Software

[25.03.04] One paper accepted at EuroS&P 2025 from Professor Simon S Woo's (DASH Lab)

The Data-driven AI & Security HCI Lab (DASH Lab, Advisor: Simon S. Woo) has had one System of Knowledge (SoK) paper accepted for publication at the 10th IEEE European Symposium on Security and Privacy (Euro S&P), a prestigious international conference covers Machine Learning Security, System & Network Security, Cryptographic Protocols, Data Privacy. The papers will be presented in July in Venice, Italy. SoK: Systematization and Benchmarking of Deepfake Detectors in a Unified Framework, EuroS&P 2025 Authors: Binh Le and Jiwon Kim (Ph.D. Student, Department of Software), Simon S. Woo (Professor, Department of Software, Sungkyunkwan University) This work is jointly performed with CSIRO Data61 as an international collaboration. Paper Link: https://arxiv.org/abs/2401.04364
- 작성일 2025-03-04
- 조회수 141
[25.03.04] 2024 SKKU Fellowship 10 Professors Selected(Professor Hyungsik Kim of the College of Software Convergence)

2024 SKKU Fellowship 10 Professors Selected Our university has selected Professor Minwoo Kim of the College of Social Sciences, Professor Ahyoung Seo of the College of Business Administration, Professor Donghee Son of the College of Information and Communications, Professor Jaehyuk Choi of the College of Information and Communications, Professor Hyungsik Kim of the College of Software Convergence, Professor Jooyoung Shin of the College of Pharmacy, Professor Jaeyeol Cho of the College of Life Sciences, Professor Sehoon Lee of the College of Medicine, Professor Youngmin Kim of the Sungkyunkwan University Convergence Institute, and Professor Honghee Won of the Samsung Institute of Convergence Medicine and Science as the '2024 SKKU-Fellowship' professors. The SKKU-Fellowship system is the highest honor that our university has awarded since 2004, and is a system that selects the best professors whose research capabilities or industry-university cooperation achievements have reached world-class standards or are highly accessible, and grants them exceptional research support and honor. The 2024 SKKU-Fellowship is based on the university management policy of “Inspiring Future, Grand Challenge” for the 23rd and 24th academic years, and selected recipients in the fields of renowned international conferences, top journals and papers, and industry-academia cooperation ecosystems by expanding the excellence and scope of each professor. The awards ceremony was held at the general faculty meeting held on Wednesday, February 19th last year. This year, the 2024 SKKU Fellowship officially recommended candidates through the Fellowship Advisory Board, which advised on the selection of candidates, and Director of Industry-Academia Cooperation Gu Ja-chun, a member of the Fellowship Advisory Board, personally announced the 10 recipients. ▲ (From the top left) Professor Kim Min-woo, Professor Seo Ah-young, Professor Son Dong-hee, Professor Choi Jae-hyeok, Professor Kim Hyung-sik, Professor Shin Joo-young, Professor Cho Jae-yeol, Professor Kim Young-min Professor Kim Min-woo of the College of Social Sciences and Professor Kim Young-min of the Sungkyunkwan University of Convergence gave their acceptance speeches on behalf of the awardees. In the future, our university will continue to discover the various excellent achievements and values of the best professors and move forward to become a first-class university that contributes to human society.
- 작성일 2025-03-04
- 조회수 176
[25.02.11] Professor Woo Hongwook’s Research Lab (CSI Lab), ICLR 2025 Paper Acceptance

[Professor Woo Hongwook’s Research Lab (CSI Lab), ICLR 2025 Paper Acceptance] Two papers from CSI Lab (Supervised by Professor Woo Hongwook) have been accepted for presentation at ICLR 2025 (The 13th International Conference on Learning Representations), a prestigious conference in the field of Artificial Intelligence. The papers will be presented in April 2025 at the Singapore Expo in Singapore. 1. Paper “Model Risk-sensitive Offline Reinforcement Learning” The author of this paper is Kwangpyo Yoo, a Ph.D. candidate in the Department of Software. This study proposes a Model Risk-sensitive Reinforcement Learning (Model Risk-sensitive RL) framework for critical mission domains, such as robotics and finance, where decision-making is crucial. The paper particularly details a model risk-sensitive offline reinforcement learning technique (MR-IQN). MR-IQN aims to minimize the "model risk" loss in cases where the model's learned data differs from the real environment, leading to decreased accuracy. To achieve this, it calculates the model's confidence in each data point and evaluates the model risk per data point using a Critic-Ensemble Criterion. It also introduces a Fourier Feature Network that limits the gap between the actual policy's value function and the inferred policy’s value in an offline setting. MR-IQN outperformed other state-of-the-art risk-sensitive reinforcement learning techniques in experiments conducted in MT-Sim (financial trading environment) and AirSim (autonomous driving simulator), achieving lower risk and higher average performance. 2. Paper “NeSyC: A Neuro-symbolic Continual Learner For Complex Embodied Tasks In Open Domains” This paper was co-authored by Wonje Choi (Ph.D. candidate, Department of Software), Jinwoo Park (Master’s student, Department of Artificial Intelligence), Sanghyun Ahn (Master’s student, Department of Software), and Daehui Lee (Integrated Master’s and Ph.D. candidate). The study proposes a Neuro-symbolic Continual Learner (NeSyC) framework that continuously generalizes knowledge (Actionable Knowledge) from embodied experiences to be applied to various tasks in open-domain physical environments. NeSyC mimics the human cognitive process of hypothesizing and deducing (hypothetico-deductive reasoning) to improve performance in open domains. This is achieved by: Using LLMs and symbolic tools to repeatedly generate and verify hypotheses from acquired experiences in a contrastive generality improvement approach. Utilizing memory-based monitoring to detect action errors of embodied agents in real-time and refine their knowledge, ultimately improving the agent's task performance and generalization across open-domain environments. NeSyC was evaluated across various benchmark environments, including ALFWorld, VirtualHome, Minecraft, RLBench, and real-world robotic tabletop scenarios. It demonstrated robust performance across dynamic open-domain environments and outperformed state-of-the-art methods, such as AutoGen, ReAct, and CLMASP, in task success rates. CSI Lab conducts research on network and cloud system optimization, autonomous driving of robots and drones, and other self-learning technologies by leveraging Embodied Agent, Reinforcement Learning, and Self-Learning. Contact Information:Professor Woo Hongwook | hwoo@skku.edu | CSI Lab | https://sites.google.com/view/csi-agent-group
- 작성일 2025-02-11
- 조회수 358
[25.02.07] 2024 SW Talent Festival Grand Prize RISE Team (CSE '23): Interview with Team Leader Jung Gi-yong

[25.02.07] 2024 SW Talent Festival Grand Prize RISE Team (Department of Computer Science and Engineering '23): Interview with Team Leader Jung Gi-yong From left: Jeong Hee-seong, Jung Gi-yong, Lee Gyu-min, Lee Sang-yeop, Lee Sang-jun On December 5th and 6th, 2024, the 2024 SW Talent Festival was held over two days. The event was hosted by the Ministry of Science and ICT and organized by the Information and Communication Planning and Evaluation Institute and the SW-Centric University Council. Under the theme “An AI World Connected by Software,” the festival showcased, exhibited, and awarded the major achievements and outstanding outcomes from 58 SW-centric universities. At this festival, the RISE team—composed of five students from the Department of Computer Science and Engineering at Sungkyunkwan University—won the Grand Prize (Minister of Science and ICT Award) by enhancing chart recognition performance through the construction of a Korean chart learning dataset. Let’s meet team leader Jung Gi-yong from the RISE team. Q: What motivated you to participate in the 2024 SW Talent Festival? A: In our Department of Computer Science and Engineering, there is a one-year industry–academia collaboration project. Out of roughly 20 teams that complete this program, professors select the most promising teams through meetings and offer them the opportunity to represent the university in competitions. My professor suggested that we participate, and that’s how we entered the contest under the name “RISE” (which carries the meanings “to awaken” and “to soar”). Q: The RISE team won the Grand Prize with “ChartBrain.AI.” Could you please explain what ChartBrain.AI is? A: ChartBrain.AI is a compact AI model that converts chart images into table text. When we started the project in April, we noticed that although the GPT-4o model understood general photographs and images quite well, it struggled with chart images—it lacked the ability to accurately extract numerical data. To address this shortcoming, we set out to develop a small AI module that converts chart images into table text in a format that can be easily understood by a Large Language Model (LLM). Because cloud-based GPT-4o models carry the risk of data leakage, companies typically do not use them for processing internal reports and chart images that require high security; instead, they deploy their own in-house LLMs. Our ChartBrain.AI is well suited for such applications. It is compact and, among domestic models, has achieved state-of-the-art (SOTA) accuracy at its current stage. Q: Could you describe the process of creating ChartBrain.AI? A: We began by taking an English model called Deplot—released by Google Research—and performed initial training to enable it to understand Korean. Then, we built a dataset of 1.12 million chart-to-table data pairs and further trained the Deplot model with this data to complete our system. Out of these, 320,000 pairs were synthetic chart images that our team created to supplement the limited diversity of existing open-source chart image datasets and to enable the model to understand even more complex charts. Note: LLM (Large Language Model): A language model built from neural networks with an enormous number of parameters. Deplot Official Code Link: https://github.com/google-research/google-research/tree/master/deplot Q: I heard that at your university’s booth you explained your award-winning work while wearing the Cheonggeumbok—the traditional attire once worn by Sungkyunkwan students. What prompted you to wear it, and what are your impressions? A: We didn’t prepare the Cheonggeumbok ourselves. The faculty member in charge of the Software Convergence College advised us to wear it during our presentation at the booth. Honestly, I felt a bit self-conscious walking around in it at first, but later, after seeing photos where it was immediately obvious that we were Sungkyunkwan University students, I grew to appreciate it. It seems our professors had remarkable foresight. Q: As the team leader, what was most important to you during the project? A: In an industry–academia collaboration project, the topic is provided by a company and the final deliverable must be submitted to them. This setup creates a greater sense of responsibility compared to typical school assignments—I constantly thought, “If we don’t deliver a proper result, it’ll be a big problem.” No matter how hard we worked, if the final program’s performance was poor, all our efforts would have been in vain. We had to keep pushing until we achieved the desired outcome. Although the process wasn’t easy, I believe that our relentless effort ultimately led us to produce excellent results. Q: What were some of the challenges you faced in submitting your program for the festival, and how did you overcome them? A: Everything was completely new to us. During the summer break, we participated in an “Industry–Academia Summer Intensive Work Program.” We rented a classroom and, much like interns, worked there from 9 a.m. to 6 p.m. every weekday. We dedicated ourselves to technical development and reading research papers. Through this process, we experienced for the first time the full cycle of reading papers, experimenting with prior research, setting a research direction based on experimental results, forming hypotheses, training models, reviewing results, and identifying shortcomings. As the team leader, I felt an even greater sense of responsibility. I believe our advisor’s active guidance and the enthusiastic participation of all team members were key to our success. Without our advisor, it would have been extremely difficult to succeed with this project. Q: Were there any particularly memorable moments during the project’s progress? A: We did not win a major award from the start. Before receiving the Grand Prize, we participated in two other competitions but were eliminated in the first round in both cases. When we entered the first competition, everyone worked incredibly hard—even staying up until 1 a.m. (the dormitory curfew) to continue development. For the second competition, we made further improvements over our previous version. By the time we entered this final competition, our morale was quite low compared to the first attempt; however, it’s very gratifying that our work eventually shone through. Q: You entered several competitions with the same project. Did the performance of the program improve significantly over time? A: Yes, there were significant improvements in performance. As mentioned earlier, our benchmark was the GPT-4 Omni model. While both our ChartBrain.AI and GPT-4 Omni were evolving simultaneously, during the summer break we were confident in stating that our model was superior. However, by December GPT-4 Omni had caught up with us for a period. In the end, our model advanced considerably and regained a comparative edge. This improvement was crucial in helping us win the Grand Prize. Q: Since the RISE team is composed entirely of CSE students, were there any classes or extracurricular activities that helped with this project? A: For me, participating in academic societies was extremely helpful. I joined an AI society called “TNT” at our university and took part in paper study sessions. As a sophomore, I was able to read many research papers, review them, and ask questions in TNT, which taught me how to discern a good paper from a less effective one. Personally, the paper review sessions in TNT were the most beneficial. Q: What do you find most attractive about studying software? A: Ultimately, software is about programming to create documents. In any company, the work I do might involve editing just a word or two in countless documents—and yet, those small changes can have a tremendous impact. Since text can be easily reproduced, even a small idea that improves one part can have infinite influence. I think that is the greatest appeal of software and what continuously fuels my competitive spirit. Q: What are your future career goals or aspirations? A: In the short term, I plan to write an undergraduate paper based on our award-winning project around May or June next year. When I work on development, I feel that roughly 50% of the help comes from GPT and about 30% from other online sources, which sometimes leaves me with the impression that I haven’t fully grasped everything. Therefore, my long-term goal is to pursue graduate studies so that I can deepen my knowledge in mathematics, English, and algorithms. Q: Do you have any advice for students preparing for software-related competitions? A: I believe that students preparing for competitions are already incredibly diligent, so here’s a tip rather than just advice. As you work on your project, you might find yourself deeply attached to your ideas. Explaining these ideas to others, especially to the judges, can be very challenging. This was the aspect that hindered me the most during competitions. I spent a lot of time thinking about how to effectively communicate my beloved idea, particularly to the judges. In my experience, effective persuasion can be achieved within just one or two PowerPoint slides. Focus on that key part, and I hope you achieve excellent results. Convince everyone possible, and I wish you great success.
- 작성일 2025-02-07
- 조회수 366
[25.01.23] IIS Lab, Four Papers Accepted at NAACL 2025

[25.01.23] IIS Lab, Four Papers Accepted at NAACL 2025 The Information and Intelligent Systems Lab (IIS Lab), led by Professor Ji-Hyung Lee, has had four papers accepted at NAACL 2025 (2025 Annual Conference of the North American Chapter of the Association for Computational Linguistics), one of the top-tier international conferences in natural language processing (NLP). The papers will be presented in April 2025 in New Mexico, USA. 1. DeCAP: Context-Aware Prompt Generation for Debiased Zero-shot Question Answering in Large Language Models (NAACL 2025) Authors: Sooyoung Bae (PhD Student, Department of Artificial Intelligence) Yoonseok Choi (Assistant Professor, Department of Computer Education, Sungkyunkwan University / PhD Graduate, Department of Software) Large language models (LLMs) perform well in zero-shot question answering (QA) tasks. However, existing methods suffer from performance gaps between ambiguous and clear questions and low debiasing performance due to strong dependence on provided instructions or internal knowledge. To address these issues, we propose DeCAP (Context-Aware Prompt Generation), which: Utilizes a Question Ambiguity Detector to reduce performance gaps caused by ambiguous question types. Employs a Neutral Next Sentence Generator to decrease dependency on internal biased knowledge by providing neutral contextual information. Experiments on BBQ and UNQOVER datasets across six LLMs show that DeCAP achieves state-of-the-art debiasing performance in QA tasks, significantly enhancing the fairness and accuracy of LLMs across diverse QA environments. 2. SALAD: Improving Robustness and Generalization through Contrastive Learning with Structure-Aware and LLM-Driven Augmented Data (NAACL 2025) Authors: Sooyoung Bae (PhD Student, Department of Artificial Intelligence) Hyojun Kim (SKT / MS Graduate, Department of Artificial Intelligence) Yoonseok Choi (Assistant Professor, Department of Computer Education, Sungkyunkwan University / PhD Graduate, Department of Software) This paper introduces SALAD (Structure-Aware and LLM-driven Augmented Data), a novel approach aimed at enhancing robustness and generalization in NLP models using contrastive learning. SALAD generates: Structure-aware positive samples using a tagging-based method. Counterfactual negative samples with diverse sentence patterns generated by LLMs. This allows the model to learn structural relationships between key sentence components while minimizing reliance on spurious correlations. We evaluate SALAD on three tasks: Sentiment Classification Sexism Detection Natural Language Inference (NLI) Results show that SALAD improves robustness and performance across different settings, including out-of-distribution datasets and cross-domain scenarios. 3. CoRAC: Integrating Selective API Document Retrieval with Question Semantic Intent for Code Question Answering (NAACL 2025) Authors: Yoonseok Choi (Assistant Professor, Department of Computer Education, Sungkyunkwan University / PhD Graduate, Department of Software) Cheolwon Na (Integrated MS/PhD Program, Department of Artificial Intelligence) Automated Code Question Answering (AQA) aims to generate precise answers for code-related queries by analyzing code snippets. However, in real-world settings, users often provide only partial code, making it difficult to derive correct answers. To address this challenge, we propose CoRAC, a knowledge-driven framework that improves AQA by: Selective API document retrieval Question semantic intent clustering We evaluate CoRAC on three real-world benchmark datasets, demonstrating its effectiveness. Results show that CoRAC generates high-quality answers outperforming LLM-based solutions like ChatGPT. 4. Q-FAKER: Query-free Hard Black-box Attack via Controlled Generation (NAACL Findings 2025) Authors: Cheolwon Na (Integrated MS/PhD Program, Department of Artificial Intelligence) Yoonseok Choi (Assistant Professor, Department of Computer Education, Sungkyunkwan University / PhD Graduate, Department of Software) Adversarial attack methods for testing language model vulnerabilities often require multiple queries and access to target model information. Even black-box attacks typically depend on target model output data, making them impractical in hard black-box settings where access is restricted. Existing hard black-box attack methods still demand high query counts and expensive adversarial generator training costs. To solve this, we introduce Q-FAKER (Query-free Hard Black-box Attacker), an efficient adversarial example generation method that: Uses a surrogate model to generate adversarial sentences without accessing the target model. Leverages controlled generation techniques for adversarial text generation. We evaluate Q-FAKER across eight datasets, demonstrating its high transferability and effectiveness in hard black-box attack scenarios. Contact Information Professor Ji-Hyung Lee | john@skku.edu IIS Lab | https://iislab.skku.edu/
- 작성일 2025-02-04
- 조회수 333
[25.01.21] Security Engineering Lab, Two Papers Accepted at CHI 2025

[25.01.21] Security Engineering Lab, Two Papers Accepted at CHI 2025 The Security Engineering Lab (Advisor: Professor Hyungsik Kim) has had two papers accepted at CHI 2025 (ACM SIGCHI Conference on Human Factors in Computing Systems), one of the top-tier conferences in the field of Human-Computer Interaction (HCI). The papers will be presented in April 2025 in Yokohama, Japan. 1. Paper: "Understanding and Improving User Adoption and Security Awareness in Password Checkup Services" Authors: Sanghak Oh (PhD Student, Department of Electrical and Computer Engineering) Heewon Baek (MS Student, Department of Electrical and Computer Engineering) Taeyoung Kim (PhD Student, Department of Electrical and Computer Engineering) Woojin Jeon (PhD Student, Department of Electrical and Computer Engineering) Junho Heo (Samsung Research) Professor Ian Oakley (KAIST) Professor Hyungsik Kim (Sungkyunkwan University) Password Checkup Services (PCS) help users protect accounts by identifying compromised, reused, or weak passwords. However, these services have low adoption rates. This study conducted an online survey (N=238) to identify factors influencing PCS adoption and barriers to changing compromised passwords. Key findings include: Adoption factors: Perceived usefulness, ease of use, and self-efficacy were significant motivators. Barriers to password changes: Warning fatigue from frequent alerts, low awareness of password compromise risks, and reliance on other security measures discouraged users from taking action. To address these issues, the research team redesigned the PCS interface by: Clarifying warning messages related to compromised passwords. Automating the password change process, such as enabling users to update multiple reused passwords simultaneously or directly linking to password change pages. A task-based interview study (N=50) validated the effectiveness of the new design, showing a significant increase in password change rates in two scenarios: 40% and 74% change rates, compared to 16% and 60% in Google's existing PCS design. 2. Paper: "I Was Told to Install the Antivirus App, but I’m Not Sure I Need It: Understanding the Adoption, Discontinuation, and Non-Use of Smartphone Antivirus Software in South Korea" Authors: Seyoung Jin (MS Student, Department of Software) Heewon Baek (MS Student, Department of Software) Professor Euijin Lee (KAIST) Professor Hyungsik Kim (Sungkyunkwan University) This study investigates the limited effectiveness of smartphone antivirus software, despite recommendations from security firms, due to user misconceptions, regulatory requirements, and improper usage. Using a mixed-methods approach, including in-depth interviews (N=23) and a survey (N=250), the study examined the adoption status of smartphone antivirus software, particularly in South Korea, where it is often mandatory for banking and financial apps. Key findings: Many users confused antivirus software with general security tools and were unaware of its limited scope in addressing mobile malware threats. Factors influencing adoption: Perceived vulnerability, response efficacy, self-efficacy, social norms, and awareness. Factors leading to discontinuation or non-use: Concerns about system performance impact and skepticism about necessity. Additionally, the mandatory installation of antivirus software for financial apps in South Korea has contributed to user misconceptions, negative perceptions, and a false sense of security. This research highlights the need for better user education, clearer communication on mobile-specific security threats, and improved guidance to enhance effective antivirus software usage.
- 작성일 2025-02-04
- 조회수 376
[24.12.26]Eunseok Ryu, Head of the Department of Immersive Media Engineering, Receives Minister of Science and ICT Award

Professor Eunseok Ryu, Head of the Department of Immersive Media Engineering, Receives Minister of Science and ICT Award Professor Eunseok Ryu, Head of the Department of Immersive Media Engineering, was awarded the Minister of Science and ICT Commendation on Wednesday, December 11, in recognition of his contributions to nurturing key talent in the metaverse field, international collaboration, and standardization efforts for next-generation technology development. Since the inauguration of the Department of Immersive Media Engineering in the second semester of 2023, Professor Ryu has been dedicated to researching and developing immersive media content technologies based on core technologies such as image processing, computer graphics, and artificial intelligence. The department has been actively selecting outstanding full-time graduate students with support from the Metaverse Convergence Graduate School Program, funded by the Ministry of Science and ICT. Additionally, the Department of Immersive Media Engineering operates an ICT and content convergence curriculum. It provides: Internship opportunities for all students in the department Overseas research institute placements for 25% of students Through these initiatives, the department is committed to fostering future global leaders in immersive media technology.
- 작성일 2025-02-04
- 조회수 287
[24.12.23] CSI Lab (Prof. Hongwook Woo), Paper Accepted at AAAI 2025

[Prof. Hongwook Woo] CSI Lab , Paper Accepted at AAAI 2025 The CSI Lab (Advisor: Professor Hongwook Woo) has had its paper accepted at AAAI 2025 (The 39th Annual AAAI Conference on Artificial Intelligence), one of the prestigious conferences in the field of artificial intelligence. The paper is scheduled to be presented in February 2025 in Philadelphia, USA. Paper Details The paper, “In-Context Policy Adaptation Via Cross-Domain Skill Diffusion,” was authored by Minjong Yoo (Integrated MS/PhD Program, Department of Software) as the first author, with Wookyung Kim (Integrated MS/PhD Program, Department of Software) as a co-author. This research proposes an In-Context Policy Adaptation (ICPAD) framework for long-horizon, multi-task environments across various domains and introduces diffusion-based skill learning techniques in cross-domain settings. ICPAD is designed to rapidly adapt reinforcement learning (RL) policies to diverse target domains using only limited target domain data—without requiring model updates. To achieve this, ICPAD: Learns domain-invariant prototype skills and domain-grounded skill adapters to maintain consistency across domains while adapting policies to new target domains through cross-domain skill diffusion. Optimizes diffusion-based skill translation by utilizing limited target domain data as prompts, enhancing policy adjustment via dynamic domain prompting. Experimental Results Experiments demonstrated that ICPAD outperforms state-of-the-art (SOTA) methods in adapting to dynamic environmental changes and various domain settings in: MetaWorld (robotic manipulation environment) CARLA (autonomous driving simulator) CSI Lab Research and Funding The CSI Lab focuses on machine learning, reinforcement learning, and self-supervised learning for optimizing networks, cloud systems, robotics, and autonomous drone navigation. This AAAI 2025 research is supported by: Core AI Technology Project for Human-Centered AI (IITP) National Research Foundation of Korea (NRF) Individual Basic Research Program Graduate School of AI ICT Elite Talent Development Program BK21 FOUR Program (BK21) Institute for Information & Communications Technology Planning & Evaluation (IITP) Samsung Electronics Contact Information Professor Hongwook Woo | hwoo@skku.edu CSI Lab | https://sites.google.com/view/csi-agent-group
- 작성일 2025-02-04
- 조회수 295
[24.12.13] CSI Lab(Prof. Hongwook Woo), Excellence & Encouragement Awards at 2024 SKKU Graduate Student Paper Awards

[Prof. Hongwook Woo] CSI Lab, Excellence & Encouragement Awards at 2024 SKKU Graduate Student Paper Awards ■ 2024 Science and Engineering Field 1. Excellence Award Category: ICT Title: LLM-based Skill Diffusion for Zero-shot Policy Adaptation Recipient: Wookyung Kim (Department of Software) 2. Encouragement Award Category: ICT Title: Exploratory Retrieval-Augmented Planning for Continual Embodied Instruction Following Recipient: Minjong Yoo (Department of Software)
- 작성일 2025-02-04
- 조회수 213
[24.12.02] LearnData Lab, Research on Graph Neural Networks Accepted at WSDM 2025

LearnData Lab's Research on Graph Neural Networks Accepted at WSDM 2025 (Master's Graduate: Jongwon Park, PhD Candidate: Heesoo Jeong) A research paper from LearnData Lab (Advisor: Professor Hogun Park) has been accepted at The 18th ACM International Conference on Web Search and Data Mining (WSDM 2025), one of the top-tier conferences in the field of artificial intelligence. Paper Details The paper, “CIMAGE: Exploiting the Conditional Independence in Masked Graph Auto-encoders”, was published with Jongwon Park (Master’s Graduate, AI Department) as the first author, and Heesoo Jeong (PhD Candidate, Software Department) as the co-first author. Research Highlights Professor Hogun Park's research team at Sungkyunkwan University has achieved significant advancements in Graph Neural Network (GNN) learning based on self-supervised learning. This study introduces a novel model called CIMAGE (Conditional Independence Aware Masked Graph Auto-Encoder), which overcomes the limitations of conventional random masking techniques and significantly enhances the expressive power of GNNs. The CIMAGE model leverages conditional independence to design a more effective masking strategy, significantly improving both efficiency and accuracy in graph representation learning. A key aspect of this research is the use of high-confidence pseudo-labels to generate two independent contexts, enabling a novel pretext task that enhances the masking and reconstruction processes. The effectiveness of CIMAGE has been demonstrated across various graph benchmark datasets, achieving outstanding performance in downstream tasks such as node classification and link prediction. This breakthrough establishes a new standard in graph representation learning. Significance and Future Applications This research represents an important milestone in Sungkyunkwan University's commitment to innovative and pioneering research. The findings have high potential for application in graph neural networks and self-supervised learning. LearnData Lab focuses on developing cutting-edge machine learning and data mining technologies across various modalities, including graphs, natural language, sensor data, and images. The lab is also actively involved in explainable AI research. The WSDM 2025 paper was supported by funding from the Graduate School of AI, the Institute for Information & Communications Technology Planning & Evaluation (IITP), and the Korea Creative Content Agency (KOCCA). Contact Information Professor Hogun Park | hogunpark@skku.edu LearnData Lab | https://learndatalab.github.io
- 작성일 2025-02-04
- 조회수 215

Center Updates