Extension of PHATE to allow data fusion Extension of PHATE to allow data fusion

The PHATE algorithm [1] is a novel tool that allows for dimensionality reduction with the benefit of conserving both local and global structure of the data. Nonetheless, it is not fit to work with continuous patterns such as those extracted from echo measurements (flow and strain traces). In this project we’ll aim at extending the capabilities of PHATE to deal with multiple kernels, rather than just one, and validate the results with real clinical data.

Preferred skills: comfortable with Matlab; notions of machine learning.


[1] Moon, K.R., van Dijk, D., Wang, Z. et al. Visualizing structure and transitions in high-dimensional biological data. Nat Biotechnol 37, 1482–1492 (2019). https://doi.org/10.1038/s41587-019-0336-3