Machine Learning Methods for Fully Exploiting Heterogeneous, Multimodal Data
To develop novel statistical methods enabling the fusion of multiple data types collected via different experimental technologies.
Results & Resources
This group published four papers on improving computational scalability, multiple hypothesis testing, and measurement linkage. They also developed a new computational method, BARcode DEmixing through Non-negative Spatial Regression (Bardensr), to meet the demand for a high spatial density of transcription required for quality imaging resolution. Together, these tools provide scalability and improved integration of multimodal data.