Greeting people of LISA,
Tomorrow we have Yann Dauphin practising his NIPS oral. Be there. NOTE: new time.
When: 13h30 Wednesday Nov. 23.
Where: TBA -- will let you know when I find out (hoping for AA3195).
The title is "The Manifold Tangent Classifier". Maybe you could also mention it's a practice talk for NIPS.
Abstract:
We
combine three important ideas present in previous work for building
classifiers: the semi-supervised hypothesis (the input distribution
contains information about the classifier), the unsupervised manifold
hypothesis (data density concentrates near low-dimensional manifolds),
and the manifold hypothesis for classification (different classes
correspond to disjoint manifolds separated by low density). We exploit a
new algorithm for capturing manifold structure (high-order contractive
autoencoders) and we show how it builds a topological atlas of charts,
each chart being characterized by the principal singular vectors of the
Jacobian of a representation mapping. This representation learning
algorithm can be stacked to yield a deep architecture, and we combine it
with a domain knowledge-free version of the TangentProp algorithm to
encourage the classifier to be insensitive to local directions changes
along the manifold. Record-breaking results are obtained and we find
that the learned tangent directions are very meaningful.
Cheers,
Aaron
--
Aaron C. Courville
Département d’Informatique et
de recherche opérationnelle
Université de Montréal
email:Aaron.Courville@gmail.com