Harri Valpolan vanhat työsivut

Olen palannut ICS:iin 1.12.2010 alkaen ja minulla on uusi työsivu. SOlin toisessa TKK:n laboratoriossa LCE:ssä.1.11.2004-30.11.2010. Vanha työsivunini.

Aiempia kiinnostuksen kohteita

• Ohjaamaton oppiminen (teoreettiset, käytännölliset ja biologiset näkökohdat)
• Bayesiläinen mallintaminen

Väittelin vuoden 2000 lopulla. Väitöskirjani käsittelee "ensembleoppiminen" -nimisen bayesiläisen menetelmän soveltamista ohjaamattomaan oppimiseen, erityisesti epälineaariseen faktorianalyysiin.

Olen soveltanut ensembleoppimista moniin generatiivisiin malleihin, ICA:n ja epälineaarisen faktorianalyysin lisäksi mm. harvakoodauksen. Bayes-ryhmän kotisivulta löytyy mm. aiheeseen liittyviä Matlab-ohjelmistoja.

Tutkimukseni taustaa.

Julkaisuluettelo

H. Valpola. (2004)
Behaviourally meaningful representations from normalisation and context-guided denoising, AI Lab, tekninen raportti, Zürichin yliopisto.

H. Valpola ja J. Särelä. (2004)
Accurate, fast and stable denoising source separation algorithms. In Proceedings of the 5th International Conference on Independent Component Analysis and Blind Signal Separation, ICA 2004, Granada, Espanja, ss. 65-72.

A. Honkela, S. Harmeling, L. Lundqvist ja H. Valpola. (2004)
Using kernel PCA for initialisation of variational Bayesian nonlinear blind source separation method. In Proceedings of the 5th International Conference on Independent Component Analysis and Blind Signal Separation, ICA 2004, Granada, Espanja, ss. 65-72.

A. Honkela ja H. Valpola. (2004)
Variational Learning and Bits-Back Coding: an Information Theoretic View to Bayesian Learning, IEEE Transactions on Neural Networks, 15(4), ss. 800-810.

A. Ilin, H. Valpola ja E. Oja. (2004)
Nonlinear Dynamical Factor Analysis for State Change Detection, IEEE Transactions on Neural Networks, 15(3), ss. 559-575.

H. Valpola, M. Harva ja J. Karhunen. (2004)
Hierarchical Models of Variance Sources, Signal Processing, 84(2), ss. 267-282.

T. Raiko, H. Valpola, T. Östman ja J. Karhunen. (2003)
Missing Values in Hierarchical Nonlinear Factor Analysis. In Proceedings of ICANN/ICONIP'03, Istanbul, Turkki, ss. 185-189.
Artikkeli: Post Script -versio, PDF version
Posteri: Post Script -versio, PDF-versio

H. Valpola, M. Harva ja J. Karhunen. (2003)
Hierarchical Models of Variance Sources. In Proceedings of the 4th International Symposium on Independent Component Analysis and Blind Signal Separation, ICA 2003, Nara, Japani, ss. 83-88.
PDF-versio (627 kb)

H. Valpola, T. Östman and J. Karhunen. (2003)
Nonlinear Independent Factor Analysis by Hierarchical Models. In Proceedings of the 4th International Symposium on Independent Component Analysis and Blind Signal Separation, ICA 2003, Nara, Japani, ss. 257-262.
PDF-versio (1 Mb)

A. Honkela ja H. Valpola. (2003)
On-line Variational Bayesian Learning. In Proceedings of the 4th International Symposium on Independent Component Analysis and Blind Signal Separation, ICA 2003, Nara, Japani, ss. 803-808.
PostScript -versio (160 kB), PDF-versio (140 kB)

A. Ilin ja H. Valpola. (2003)
On the Effect of the Form of the Posterior Approximation in Variational Learning of ICA Models. In Proceedings of the 4th International Symposium on Independent Component Analysis and Blind Signal Separation, ICA 2003, Nara, Japani, ss. 915-920.
Post Script -versio, PDF-versio

A. Honkela, H. Valpola ja J. Karhunen. (2003)
Accelerating Cyclic Update Algorithms for Parameter Estimation by Pattern Searches, Neural Processing Letters, 17(2), ss. 191-203.

M. Funaro, E. Oja ja H. Valpola. (2003)
Independent component analysis for artefact separation in astrophysical images, Neural Networks, 16(3-4), ss. 469-478.
PDF-versio saatavana ElsevierComputerScience:lta

H. Valpola, E. Oja, A. Ilin, A. Honkela ja J. Karhunen. (2003)
Nonlinear blind source separation by variational Bayesian learning, IEICE Trans. on Fundamentals of Electronics, Communications, and Computer Sciences, E86-A(3), ss. 532-541.
PDF-versio

H. Valpola, A. Honkela ja J. Karhunen. (2002)
An Ensemble Learning Approach to Nonlinear Dynamic Blind Source Separation Using State-Space Models. In Proceedings of the International Joint Conference on Neural Networks (IJCNN'02), Honolulu, Havaiji, USA, ss. 460-465.
Post Script -versio (241 kB), PDF-versio (308 kB)

H. Valpola ja J. Karhunen. (2002) An Unsupervised Ensemble Learning Method for Nonlinear Dynamic State-Space Models. Neural Computation 14(11), ss. 2647-2692, MIT Press.
Post Script -versio (654 kb) ja PDF-versio (937 kb)
Matlab-koodi Bayes-ryhmän sivulla.

H. Valpola, T. Raiko ja J. Karhunen. (2001) Building Blocks for Hierarchical Latent Variable Models. In Proceedings of the 3rd International Conference on Independent Component Analysis and Blind Signal Separation, ICA 2001, San Diego, Kalifornia, USA, ss. 716-721.
Post Script -versio (48 kb)

A. Iline, H. Valpola ja E. Oja. (2001) Detecting Process State Changes by Nonlinear Blind Source Separation. In Proceedings of the 3rd International Conference on Independent Component Analysis and Blind Signal Separation, ICA 2001, San Diego, Kalifornia, USA, ss. 710-715.
Post Script -versio (152 kb) ja PDF-versio (465 kb)

J. Särelä, H. Valpola, R. Vigário ja E. Oja. (2001) Dynamical Factor Analysis of Rhythmic Magnetoencephalographic Activity. In Proceedings of the 3rd International Conference on Independent Component Analysis and Blind Signal Separation, ICA 2001, San Diego, Kalifornia, USA, ss. 457-462.

M. Funaro, E. Oja ja H. Valpola. (2001) Artefact Detection in Astrophysical Image Data Using Independent Component Analysis. In Proceedings of the 3rd International Conference on Independent Component Analysis and Blind Signal Separation, ICA 2001, San Diego, Kalifornia, USA, ss. 49-54.

T. Raiko ja H. Valpola. (2001) Missing Values in Nonlinear Factor Analysis. In Proceedings of the 8th International Conference on Neural Information Processing, ICONIP 2001, Shanghai, Kiina, ss. 822-827.
Post Script -versio (292 kb)

K. Lagus, E. Alhoniemi ja H. Valpola. (2001) Independent Variable Group Analysis. In Proceedings of International Conference on Artificial Neural Networks - ICANN 2001, Wien, Itävalta, ss. 203-210.
Post Script -versio (69 kb)

H. Valpola, A. Honkela ja J. Karhunen. (2001) Nonlinear Static and Dynamic Blind Source Separation Using Ensemble Learning. In Proc. of the Int. Joint Conf. on Neural Networks, IJCNN'01, Washington D.C., USA, ss. 2750-2755.

H. Valpola. (2001) Learning hidden dynamical systems. Learning workshop, Snowbird, Utah, USA.

X. Giannakopoulos ja H. Valpola. (2001) Nonlinear Dynamical Factor Analysis. In Proceedings of The Twentieth International Workshop on Bayesian Inference and Maximum Entropy Methods in Science and Engineering, MaxEnt 2000, Pariisi, Ranska, ss. 305-317.
Post Script -versio (422 kb)

H. Valpola. (2000) Bayesian Ensemble Learning for Nonlinear Factor Analysis. Väitöskirja, Teknillinen korkeakoulu, Espoo. Julkaistu lehdessä Acta Polytechnica Scandinavica, Mathematics and Computing Series No. 108.
Post Script -versio (1,5 Mb)

H. Valpola. (2000) Unsupervised Learning of Nonlinear Dynamic State-Space Models, Publications in Computer and Information Science A59, Teknillinen korkeakoulu, Espoo, Suomi.
Post Script -versio (341 kb)

H. Valpola, X. Giannakopoulos, A. Honkela ja J. Karhunen. (2000) Nonlinear Independent Component Analysis Using Ensemble Learning: Experiments and Discussion. In Proceedings of the Second International Workshop on Independent Component Analysis and Blind Signal Separation, ICA 2000, Helsinki, Suomi, ss. 351-356.
Post Script -versio (386 kb)

H. Valpola. (2000) Nonlinear Independent Component Analysis Using Ensemble Learning: Theory. In Proceedings of the Second International Workshop on Independent Component Analysis and Blind Signal Separation, ICA 2000, Helsinki, Suomi, ss. 251-256.
Post Script -versio (63 kb)

H. Valpola ja P. Pajunen. (2000) Fast Algorithms for Bayesian Independent Component Analysis. In Proceedings of the Second International Workshop on Independent Component Analysis and Blind Signal Separation, ICA 2000, Helsinki, Suomi, ss. 233-237.
Post Script -versio (167 kb)

(Huom: Lappalainen = nykyisin Valpola.)

H. Lappalainen, A. Honkela, X. Giannakopoulos ja J. Karhunen. (2000) Nonlinear Source Separation Using Ensemble Learning and MLP Networks, to appear in Proceedings of the Symposium 2000 on Adaptive Systems for Signal Processing, Communications, and Control, AS-SPCC, Lake Louise, Alberta, Kanada.
Post Script -versio (367 kb)

H. Lappalainen ja A. Honkela. (2000) Bayesian Nonlinear Independent Component Analysis by Multi-Layer Perceptrons. In M. Girolami, editor, Advances in Independent Component Analysis, ss. 93-121, Springer-Verlag.
Post Script -versio (420 kb)

H. Lappalainen ja J. W. Miskin. (2000) Ensemble Learning. In M. Girolami, editor, Advances in Independent Component Analysis, ss. 75-92, Springer-Verlag.
Post Script -versio (127 kb)

H. Lappalainen. (1999) Fast Fixed-Point Algorithms for Bayesian Blind Source Separation, Publications in Computer and Information Science A56, Helsinki University of Technology, Espoo, Finland.
Post Script -versio (277 kb)

H. Lappalainen, X. Giannakopoulos ja J. Karhunen. (1999) Nonlinear Blind Source Separation Using Ensemble Learning. Esitetty Snowbird Learning Workshopissa.
Post Script -versio (38 kb)

H. Lappalainen ja X. Giannakopoulos. (1999) Multi-Layer Perceptrons as Nonlinear Generative Models for Unsupervised Learning: a Bayesian Treatment. In Proceedings of the ICANN'99, Edinburgh, Skotlanti, ss. 19-24.
Post Script -versio (126 kb)

H. Lappalainen. (1999) Ensemble Learning for Independent Component Analysis. In Proceedings of the First International Workshop on Independent Component Analysis and Blind Signal Separation, ICA'99, Aussois, Ranska, ss. 7-12.
Post Script -versio (90 kb)

H. Lappalainen. (1998) Using an MDL-based cost function with neural networks. In Proceedings of the IJCNN'98, Anchorage, ss. 2384-2389.
Kommentteja, Post Script -versio (63 kb)

T. Kohonen, S. Kaski ja H. Lappalainen. (1997) Self-organized formation of various invariant-feature filters in the Adaptive-Subspace SOM. Neural Computation, 9(6), ss. 1321-1344.
Post Script -versio (246 kb)

T. Kohonen, S. Kaski, H. Lappalainen ja J. Salojärvi. (1997) The Adaptive-Subspace Self-Organizing Map (ASSOM). Workshop on Self-Organizing Maps (WSOM'97), Espoo, Suomi, ss. 191-196.

H. Lappalainen. (1996) Soft multiple winners for sparse feature extraction. In Proceedings of the ICNN'96, Washington D.C., USA, ss. 207-210.
Post Script -versio (196 kb)

S. Laine, H. Lappalainen ja S.-L. Jämsä-Jounela. (1995) On-line determination of ore type using cluster analysis and neural networks. Minerals Engineering, 8(6), ss. 637-648.