Speaker:
Anna Konstorum
Institution:
Center for Computing Sciences / Institute for Defense Analyses (CCS/IDA)
Time:
Monday, November 18, 2024 - 4:00pm to 4:50pm
Host:
Location:
RH 306
Tensor decompositions are being more frequently employed to discover low-dimensional features of complex datasets. In this talk, we discuss how the decomposition of immune profiling data, which has an inherent multi-index array structure, can be viewed through the lens of a tensor dictionary learning problem. We show how this approach leads to improved rank selection and interpretability of the final model.