My research lies between computer science and chem-informatics. In particular, I am interested in machine learning and computational methods with a great emphasis on their applications to molecular graphs. Traditionally, designing, synthesizing, and evaluating new substances and procedures are meticulous processes that may span years, even decades. With the increasing availability of chemical and molecular data recently, I am hoping to develop data-driven techniques, such as new machine learning methods, to aid the discovery of new materials, potentially cutting down on the development time. Because of the general nature of the methods, any SET within BioPACIFIC MIP could be benefited in one of the following ways:
Prediction (Forward Problem): Properties of a given material design (e.g. mechanical, optical, or chemical response) are predicted in advance of fabrication. At the moment, the accuracy of the predictions varies with the properties being considered and the availability of data. Graph neural networks are the de facto method for such task with new research directions on more expressive machine learning models (transformers) are underway. Synthesis (Inverse Problem): The inverse problem, involving the design of a material which possesses a set of desired properties, is also of central importance. The adaptation of state-of-the-art generative models (VAE and GAN) to goal-directed molecule generation has achieve significant performances although limitations still exist due to the scarcity of data. Discovery of design patterns responsible for certain properties is a closely related task that aids the inverse problem.
What excites you about NSF BioPACIFIC MIP?
The BioPACIFIC MIP community is an excellent environment for me to grow academically and professionally. Personally, I would love to be in an extended group doing interdisciplinary research. As a BioPACIFIC MIP Fellow, I would have great opportunities to interact, exchange idea, and potentially collaborate with my peers and my seniors, both within and outside of my field, whose research topics align with my own interest. Coming from computer science, it is important to me that my research is not only theoretically sound, but also meaningful in terms of its application domain, hence my interest in chem-informatics and material science. The SETs and the interdisciplinary nature of BioPACIFIC MIP fits my goal perfectly.