TY  - JOUR
AB  - Modern nonlinear dimensionality reduction (DR) techniques project
high dimensional data to low dimensions for their visual inspection.
Provided the intrinsic data dimensionality is larger than two, DR nec-
essarily faces information loss and the problem becomes ill-posed. Dis-
criminative dimensionality reduction (DiDi) offers one intuitive way
to reduce this ambiguity: it allows a practitioner to identify what is
relevant and what should be regarded as noise by means of intuitive
auxiliary information such as class labels. One powerful DiDi method
relies on a change of the data metric based on the Fisher information.
This technique has been presented for vectorial data so far. The aim
of this contribution is to extend the technique to more general data
structures which are characterised in terms of pairwise similarities only
by means of a kernelisation. We demonstrate that a computation of
the Fisher metric is possible in kernel space, and that it can efficiently
be integrated into modern DR technologies such as t-SNE or faster
Barnes-Hut-SNE. We demonstrate the performance of the approach
in a variety of benchmarks.
DA  - 2017
DO  - 10.1016/j.neucom.2017.01.104
LA  - eng
IS  - SI
M2  - 34
PY  - 2017
SN  - 0925-2312
SP  - 34-41
T2  - Neurocomputing
TI  - Efficient Kernelization of Discriminative Dimensionality Reduction
UR  - https://nbn-resolving.org/urn:nbn:de:0070-pub-29093721
Y2  - 2024-11-22T04:33:43
ER  -