Machine Learning

AI Service Functionalities

We are exploring the advantage of AI techniques for accelerating scientific discovery in material science. In particular, we are developing image processing algorithms for nano-device edge height estimation and a similar image retrieval algorithm for nano-device microscopy images.

Although these algorithms work well with themselves, deploying them in the real world and benefitting scientists is still challenging. Due to data privacy concerns, data produced by scientists may contain information crucial to their research and sharing these images. Due to this challenge, our aim is to develop decentralized AI service functionalities that can train machine learning models and make a prediction with minimal amount of data sharing.