Speaker
Kirandeep Kour
(Max Planck Institute Magdeburg)
Description
The machine learning model for binary classification for tensor input data was proposed in my previous work. The main key point was to compute the kernel matrix for each pair of tensor input data, more efficiently. Along with it, we have explained TT-CP expansion and other theoretical aspects of this model which controls the stability and reliability aspects of it.
In the paper named "Exponential Machine", authors have used linear boundary classification along with tensorized-parameter. These parameters, the so-called weight tensor is factorized into TT format. The benefit of taking TT decomposition of weight parameter is that the TT-rank of the weight tensor is a hyper-parameter of the proposed method and it controls the efficiency vs. flexibility trade-off.
I am interested in looking at combining both the results to compute nonlinear boundary (kernelized Support Tensor Machine) with factorized weight tensor parameters using TT-CP decomposition despite the TT decomposition.
Primary author
Kirandeep Kour
(Max Planck Institute Magdeburg)
Co-authors
Sergey Dolgov
(University of Bath)
Prof.
Peter Benner
(Max Planck Institute for Dynamics of Complex Technical Systems)
