The landscape of material science and manufacturing has transformed significantly with the advent of machine learning (ML). This toolkit of data driven methods accelerated the discovery and production of new materials by accurately predicting the complicated physical processes and mechanisms that are not fully described by existing material theories. Yet, with an array of intricate ML models at our disposal, the pressing question remains: Which ML algorithm is best suited for our needs? In this line of research, we aim to provide insights to strategically select appropriate models aligned with specific design challenges.

Review on strategic data driven design summarizing  existing efforts
Junhyeong Lee, Donggeun Park, Mingyu Lee, Hugon Lee, Kundo Park, Ikjin Lee, and Seunghwa Ryu* "Machine learning-based inverse design methods considering data characteristics and design space size in materials design and manufacturing: a review", Materials Horizons 10, 5436 (2023). [engrxiv] [link to paper]  

Perspective on data-efficient and user-friendly AI for future
Hugon Lee†, Hyeongbin Moon†, Junhyeong Lee†, and Seunghwa Ryu "Toward Knowledge-Guided AI for Inverse Design in Manufacturing: A Perspective on Domain, Physics, and Human-AI Synergy", submitted. [arxiv] 

Our team specializes in designing theoretical and computational models for materials at various scales. We conduct atomistic simulations to understand the mechanical behavior of materials, using these insights as inputs for larger-scale continuum and composite modeling. Additionally, we develop multiphysics simulation frameworks and integrate them into existing Finite Element Analysis (FEA) codes. These simulations are further enhanced by incorporating experimental data from our group and others. To elevate manufacturing quality, we apply computational modeling tools, including those for 3D printing processes and injection molding. Furthermore, we incorporate advanced fatigue life prediction models into commercial software, which, when combined with AI algorithms, enhances the durability of structures under cyclic loading.

Many real-world design problems in manufacturing suffer from sparse, high-cost data and complex physical constraints, making purely data-driven AI models unreliable. Physics-Informed Machine Learning (PIML) addresses this by embedding governing laws into the learning process, enabling data-efficient and physically consistent modeling even in low-data regimes.

Our recent focus has been on identifying material properties under extremely data-scarce conditions, aiming to enable high-throughput material characterization and screening for AI-based design.

1. Hyeonbin Moon†, Donggeun Park†, Jinwook Yeo, and Seunghwa Ryu* "Physics-informed neural network framework for solving forward and inverse flexoelectric problems", submitted. 

2. Hyeonbin Moon†, Songho Lee†, Wabi Demeke*, Byungki Ryu, and Seunghwa Ryu* "Physics-Informed Neural Operators for Generalizable and Label-Free Inference of Temperature-Dependent Thermoelectric Properties", submitted. [arxiv] 

3. Hyeonbin Moon†, Donggeun Park†, Hanbin Cho, Hong-Kyun Noh, Jae Hyuk Lim*, and Seunghwa Ryu* "Physics-Informed Neural Network-Based Discovery of Hyperelastic Constitutive Models from Extremely Scarce Data", submitted. [arxiv] 

Large Language Models (LLMs) enable intuitive, human-centered interaction in inverse design by translating natural language queries into actionable design tasks, simulations, or optimization workflows. Combined with tools like retrieval-augmented generation and multi-agent systems, LLMs enhance accessibility, reasoning, and collaboration in AI-driven manufacturing environments.

1. Donggeun Park, Hyeonbin Moon, and Seunghwa Ryu "A Self-Correcting Multi-Agent Framework for Language-Based Physics Simulation and Explanation", submitted.  

2. Inhyo Lee, Junhyeong Lee, Seunghwa Ryu et al. "LLM-Based Multi-Agent Framework for Knowledge-Guided Discovery of Thermodynamically Stable Double Perovskites", in preparation 

3. Junyoung Kim, Junhyeong Lee, Seunghwa Ryu et al. "LLM-Based Multi-Agent Framework with Diffusion-Aided Design Generation for Injection Molding Process Optimization", in preparation