IMPRS Seminar - Tandem Talk with Kirandeep Kour and Prof. Dr. Peter Benner
IMPRS Seminars - Tandem Talk: "Tensor methods for Machine Learning"
- Date: Jan 13, 2023
- Time: 02:00 PM - 03:30 PM (Local Time Germany)
- Speaker: Kirandeep Kour, Prof. Dr. Peter Benner
- Location: Magdeburg
- Room: Online
- Host: IMPRS
The next IMPRS Seminar will be held by Prof. Dr. Peter Benner and Kirandeep Kour
They will speak about "Low -rank Tensor Decompositions in Kernel-based Machine Learning". The talk will be held hybrid. It is possible to attend online as well as in person.
Tandem talks at IMPRS seminars are part of the IMPRS program and an excellent opportunity for researchers to exchange current findings and challenges. If you want to take part in the talks, please write a mail to imprs@mpi-magdeburg.mpg.de for login details.
Everyone interested is invited to the talks.
Abstract:
Enormous amounts of data are being generated, in several scientific
applications, e.g., neuroscience, medical science, and signal processing.
Generally, these data depend on various parameters and thus can be in-
terpreted as multidimensional data (tensor) and incorporate structure in
the multi-dimensionality. Working with tensor datasets usually requires
high computational cost due to the curse of dimensionality. The Low-rank
tensor decompositions (LRTD) reduce the aforementioned issue and aim
at providing a natural and compact structure-preserving representation.
In the doctoral project, we have worked on the development of LRTD
algorithms for binary classification with small sample-sized datasets like
functional Magnetic Resonance Imaging (fMRI), CT scans, and Hyper-
spectral Images. The kernel computation (feature extraction technique)
in the Kernel-based method such as Support Tensor Machine (extended
version of Support Vector Machine for tensor data) benefits from newly
established LRTD methods. The benefits include achieving a robust, com-
putationally efficient machine learning model along with state-of-the-art
classification accuracy.
The talk will include an introductory session, followed by a detailed
description of the work with some numerical experiments and results.