Internship 8
Host name: Adrian Buganza Tepole
School and Department: School of Engineering and Applied Sciences, Mechanical Engineering
Internship title: Digital Twins of Tissue Growth and Remodeling using Tensor Product Decomposition and Neural Networks
Number of interns to be hosted: 1 (one)
Types of support offered:
- Stipend: $1500/month for 3-6 months depending on final scope agreed by student and supervisor
- Access to Columbia University campus services, computational resources
- Mentorship and supervision by faculty and senior lab members
- Immigration and Visa assistance through Columbia’s International Students and Scholars Office (ISSO)
Internship description
Student will work in the development of new reduced order models for rapid instantiation, calibration, and evaluation of digital twins for biomedical applications. Tissues have the unique ability to grow and remodeling. My group has developed models of skin growth and remodeling in response to disease, reconstructive surgery, wound healing, and drug delivery. These models are traditionally implemented with finite element simulations. For creation of digital twins, fast model evaluation is required. This internship will explore the use of tensor products to decompose multidimensional functions, with neural network building blocks. This mode of function compression allows for efficient interpolation of the finite element solution space in real time, i.e. the surrogate needed for the digital twin applications.
Skills required
Finite element simulation knowledge; good programming skills with python; basic knowledge of machine learning (e.g. simple multi-layer perceptrons or auto encoders).
Additional information
Not required, but preference to applicants with advanced scientific machine learning and applied math knowledge.