Joint Kernel Maps
Jason Weston, Bernhard Schoelkopf, Olivier Bousquet, Tobias Mann and William Stafford Noble
We develop a methodology for solving high dimensional
dependency estimation problems between pairs of
data types, which is viable in the case where
the output of interest
has very high dimension, e.g. thousands of dimensions.
This is achieved by mapping the objects
into continuous or discrete spaces, using joint kernels.
Known correlations between input and output can be defined
by such kernels, some of
which can maintain linearity in the outputs to provide
simple (closed form) pre-images.
We provide examples of such kernels and empirical results
on mass spectrometry prediction
and mapping between images.
Supplementary Experiments (artificial data, digit reconstruction)
View full image mapping (smiling face) results