Exploring and Enjoying AI Co-Creation for Scientific Discovery

DAM Laboratory
Professor：DAM Hieu-Chi

E-mail：
［Research areas］
AI, Data Science, Materials Informatics
［Keywords］
AI, Data Mining, Materials Informatics

Skills and background we are looking for in prospective students

Basic knowledge of statistics, linear algebra, and programming.
Interest in observing real-world phenomena, making inferences, and exploring inverse problems.

What you can expect to learn in this laboratory

Students will learn how to apply AI and machine learning to scientific discovery, developing skills in data analysis, pattern recognition, and inverse problem-solving. They will explore AI-assisted hypothesis generation and gain hands-on experience in co-creating with AI to extract insights from complex data. Research will cover model development and applications, emphasizing how AI can enhance scientific reasoning and accelerate breakthroughs across diverse research domains. Through interdisciplinary exploration, students will refine their ability to integrate AI into their problem-solving processes and push the boundaries of knowledge.

【Job category of graduates】 AI research scientist, Data scientist, Technology R&D, Startups

Research outline

Our lab explores how cutting-edge artificial intelligence (AI) can accelerate and enhance the process of scientific discovery. By merging machine learning, deep learning, statistical inference, evidence theory, and optimization techniques, we develop data-driven frameworks that extract hidden insights, generate new hypotheses, and support refined scientific reasoning.
Three overarching themes guide our work:

1) AI-Driven Scientific Discovery:

We integrate computational modeling and statistical methods to investigate complex systems, identify patterns in large datasets, and propose innovative solutions or theories in scientific domains.

2) AI Co-Creation:

Our group emphasizes collaborative/co-creative intelligence--- combining human expertise with AI’s analytical power. This synergy helps to form robust scientific inquiries, design new experiments, and interpret results with deeper context and clarity.

3) Inverse Problem-Solving:

We tackle situations where the data is observed, but the underlying mechanisms or causes are unclear. Through computational and algorithmic approaches, we can infer those hidden factors and guide decision-making or design processes in diverse scientific applications.

Figure 1: Attention Mechanism for discovering knowledge on materials

Figure 2: Explainable Deep Learning Architecture for Materials Science Studies

Key publications

1. M. Q. Ha and H. C. Dam et al., “Evidence-based recommender system for high-entropy alloys”, Nature Computational Science, 1, 470–478 (2021).
2. T. S. Vu and H. C. Dam et al., “Towards understanding structure–property relations in materials with interpretable deep learning”, npj Computational Materials, 9, 215 (2023).
3. T. Isogai and H. C. Dam, “Building classification trees on Japanese stock groups partitioned by network clustering”, IEEJ Transactions on Electronics, Information and Systems, 137, 10 (2017).

Equipment

High-Performance Computing Clusters: Multiple high-end servers with multi-core CPUs and GPUs for parallel computations, simulations, and deep learning tasks.