WHAT WE DO
HANA Lab works on hybrid AI which combines the physical and chemical knowledge of chemical engineering with the power of machine learning. We focus on chemical and biological problems where data is scarce and systems are complex. We aim to build AI that knows the laws of physics and can therefore deliver accurate, explainable prediction, interpretation, and design from limited data.
APPROACH
We combine what is known from physical laws with what is learned from data in a single model. Physical knowledge provides the backbone, so the model can be trained with little data. Each part of the model has a clear meaning, which makes the results easy to interpret. Physics for what we know, data for what we don't: models we can trust even with limited data.
We find the evidence behind each prediction and present it in a form people can understand. Because the reasoning can be checked, the results can be trusted and used in practice. The relationships the AI uncovers can also lead to new scientific hypotheses. We build AI that gives not only answers but also the reasons behind them.
DOMAINS
We work to understand cell culture processes, predict the quality of medicines, and pave the way for manufacturing next-generation medicines.
We design processes for clean energy production, detect process faults early, and ultimately aim for autonomous manufacturing.
We work to understand how secondary batteries behave, monitor the health of batteries in service, and pave the way for discovering next-generation materials.
JOIN US
We are looking for researchers from chemical engineering, materials science, computer science, or a related field who want to build AI that understands physical systems.