Keynote Lectures
 |
Data-Driven Modeling with Data Fusion Technique and Its Application to Topology OptimizationProf. Ikjin Lee
Abstract
Recent advances in data-driven modeling have enabled efficient prediction and optimization of complex engineering systems by integrating heterogeneous data sources. In particular, data fusion techniques that combine information from different fidelity levels have emerged as effective tools for improving model accuracy under limited high-fidelity data. This keynote presents recent progress on data-driven modeling frameworks based on multi-fidelity data fusion and their application to topology optimization. First, a data fusion-based surrogate modeling framework is introduced, where inexpensive low-fidelity data capturing global trends are combined with sparse high-fidelity data to construct accurate predictive models. The framework incorporates adaptive sampling and fidelity-aware modeling to efficiently utilize heterogeneous datasets and reduce computational cost while maintaining prediction accuracy. This approach is particularly effective in scenarios where high-fidelity simulations are expensive and data availability is limited. The second part of the talk focuses on topology optimization. To address the sensitivity of topology optimization to algorithmic hyperparameters and the risk of local-optimum trapping, a data-driven multi-fidelity topology optimization framework is presented. Early-terminated topology optimization runs are used to generate low-fidelity data, while fully converged runs provide high-fidelity information. These datasets are fused through a multi-fidelity surrogate model defined in the hyperparameter space, enabling efficient global exploration under limited computational budgets. Numerical examples demonstrate that the proposed approach can identify improved topology designs compared with conventional methods while significantly reducing computational cost.
Biography
Ikjin Lee is a Professor in the Department of Mechanical Engineering at the Korea Advanced Institute of Science and Technology (KAIST). He received his B.S. (2001) and M.S. (2003) degrees in Mechanical Engineering from Seoul National University, and his Ph.D. degree (2008) in Mechanical Engineering from the University of Iowa, USA. After completing his Ph.D., he worked as a Postdoctoral Research Scholar and Adjunct Professor at the University of Iowa (2008–2011), and later as an Assistant Professor at the University of Connecticut (2011–2013). In 2013, he joined KAIST as an Assistant Professor, where he was promoted to Associate Professor in 2017 and Professor in 2023. Professor Lee actively contributes to the academic community, serving as President of Korea Society of Design Optimization (KSDO), Vice President of Korea Society of Computational Mechanics (KSCM), an Associate Editor for the ASME Journal of Mechanical Design (2021–present), an Associate Editor for the Journal of Mechanical Science and Technology (2020–2023), and an Editor for Transactions of the Korean Society of Mechanical Engineers A (2022–present). Professor Lee’s research interests include reliability-based and robust design optimization (RBDO/RDO), surrogate modeling and AI-driven design optimization, and sampling-based optimization and model validation. |
 |
Explicit Topology Optimization for Mechanical Metamaterials: Recent Advances and PerspectivesProf. Zongliang Du
Abstract
Mechanical metamaterials offer unprecedented opportunities to realize mechanical properties surpass those achievable with conventional materials by rationally designing architectured microstructures. Enabled by recent advances in additive manufacturing, the inverse design of mechanical metamaterials-closely aligned with topology optimization-has become a rapidly evolving frontier in mechanics and materials science. This talk presents a systematic design paradigm based on explicit topology optimization, with particular emphasis on the moving morphable components/voids (MMC/MMV) method. The proposed approach features a reduced number of design variables, crisp structural boundaries, and a natural decoupling between finite element analysis and topology representation. Several representative classes of mechanical metamaterials are optimally designed and showcased, including topological metamaterials with defect-immune energy transport, mechanical cloaks capable of concealing embedded objects, and metamaterials exhibiting programmable force–displacement responses. Furthermore, by integrating explicit topology optimization with deep learning techniques, we develop a structural genome database that enables the rapid, on-demand generation of mechanical metamaterials with tailored effective elastic properties. Finally, perspectives on emerging opportunities and future research directions in this field are discussed.
Biography
Prof. Du received his B.S. degree from Dalian University of Technology in 2009 and his Ph.D. in Engineering Mechanics from the same institution in 2016. He conducted postdoctoral research at the University of California, San Diego from 2017 to 2018 and at the University of Missouri in 2019. Since 2020, he has served as an Associate Professor in the Department of Engineering Mechanics at Dalian University of Technology, where he was promoted to Full Professor in 2024. He has also been serving as Vice Dean of the department since 2024. Prof. Du’s research focuses on structural optimization and its applications to advanced materials and structures. He has published more than 80 peer-reviewed journal papers, including articles in leading journals of mechanics such as the Journal of the Mechanics and Physics of Solids and Computer Methods in Applied Mechanics and Engineering. His work has received over 3,500 citations according to Google Scholar. He serves as an Editorial Board Member of Advances in Mechanics, and as a Young Editorial Board Member of Acta Mechanica Sinica, Computers & Structures, and Acta Mechanica Solida Sinica. He was the recipient of the Young Scientist Award at the Asian Congress of Structural and Multidisciplinary Optimization (ACSMO) in 2020. |
 |
Human-Centered Optimization in Design Engineering
|
 |
From Points to Patterns: Rethinking Optimization Landscapes using iSOMProf. Palaniappan Ramu
Abstract
human intuition. This keynote introduces a paradigm shift through iSOM, where visualization is not merely a post-processing tool, but a central driver of the optimization process itself. By transforming high-dimensional design spaces into interpretable visual structures, ISOM enables designers and engineers to see performance landscapes, uncover hidden relationships, and make informed decisions with unprecedented clarity. The talk will demonstrate how this approach fundamentally reshapes key aspects of modern optimization. We begin with Region of Interest (RoI) identification, where visualization-guided exploration allows rapid narrowing of promising design spaces, significantly reducing computational effort. Building on this, multi-objective optimization (MOO) is reframed as an interactive and intuitive process, enabling clearer trade-off navigation beyond traditional Pareto front analysis. The keynote will further explore how ISOM integrates seamlessly with emerging paradigms such as transfer learning leveraging prior knowledge across design problems and robust design, where uncertainty can be visualized and managed more effectively. Finally, the concept of graphical optimization will be presented as a unifying framework, where optimization evolves from algorithm-centric to insight-driven, empowering human-in-the-loop decision-making.
Biography
Prof. Palaniappan Ramu is a faculty at the Dept. Engineering Design, IIT Madras. He received his PhD and MS from Univ. Florida, USA and B.Eng from Madurai Kamaraj University. After brief stints at Belcan and Univ. Notre Dame he joined IIT Madras in 2010. He is currently a review editor of J. Structural and Multidisciplinary Optimization and is a founding member of the Indian community for MDAO. His research interests revolve around, uncertainity quantification, AI in Engineering Design, Data visualization and Engineering analytics. |