Aapo Hyvärinen: Publications

Publications by topic

Publications front page

Home page

Unsupervised learning: misc.

B. Sriperumbudur, K. Fukumizu, A. Gretton, A. Hyvärinen, R. Kumar. Density Estimation in Infinite Dimensional Exponential Families. J. of Machine Learning Research, 18(57):1-59, 2017.
[A new approach to density estimation combining kernel methods with score matching.]

H. Sasaki, M. Sugiyama, and A. Hyvärinen. Clustering via Mode Seeking by Direct Estimation of the Gradient of a Log-Density. Proc. European Conf. on Machine Learning (ECML2014), Nancy, France, 2014.
[A new version of mean-shift clustering, using a more sophisticated density estimator. In high dimensions, much better than the original.]

M. Gutmann and A. Hyvärinen. Learning reconstruction and prediction of natural stimuli by a population of spiking neurons Proc. European Symposium on Artificial Neural Networks (ESANN2009), Bruges, Belgium, 2009.
[Shows how to find a statistically motivated representation in a population of neurons using a biophysically motivated neuron model.]

M. Gutmann, A. Hyvärinen, and K. Aihara. Learning encoding and decoding filters for data representation with a spiking neuron Proc. Int. Joint Conf. on Neural Networks (IJCNN2008), Hong Kong, 2008.
[A predecessor of the paper above: A method similar to PCA but using a biological model of a neuron.]

A. Hyvärinen. Behavioural priors: Learning to search efficiently in action planning. Proc. European Conference on Cognitive Science (EuroCogSci07), Delphi, Greece, pp. 324--328, 2007.
[Proposes how planning can be improved by building a prior model of useful behavioural sequences.]

M.A. Vicente, P.O. Hoyer and A. Hyvärinen. Equivalence of some common linear feature extraction techniques for appearence-based object recognition tasks. IEEE Transactions on Pattern Analysis and Machine Intelligence, 29(5):896-900, 2007.
[Shows that ICA features are often equivalent, in terms of classification performance, to the features given by whitening, and discusses how ICA features should be used to improve classification.]

A. Hyvärinen and J. Perkiö. Learning to segment any random vector. Proc. Int. Joint Conf. on Neural Networks (IJCNN2006), pp. 4167-4172, Vancouver, Canada, 2006.
pdf  matlab code
[Proposes a probabilistic framework for learning to segment observations of any random vector. A generalization of models of image segmentation.]

A. Hyvärinen. Consistency of pseudolikelihood estimation of fully visible Boltzmann machines. Neural Computation, 18(10):2283-2292, 2006.
pdf  gzipped ps
[A very simple estimation method for fully visible BM's, with guaranteed consistency.]