Mathias Berglund

Doctoral Candidate

Research outputs

  1. 2019
  2. Published

    Regularizing Trajectory Optimization with Denoising Autoencoders

    Boney, R., Di Palo, N., Berglund, M., Ilin, A., Kannala, J., Rasmus, A. & Valpola, H., 2019, 33rd Conference on Neural Information Processing Systems: NeurIPS 2019 . Neural Information Processing Systems Foundation, (Advances in Neural Information Processing Systems).

    Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

  3. 2017
  4. Published

    Unsupervised Networks, Stochasticity and Optimization in Deep Learning

    Berglund, M., 2017, Aalto University. 214 p.

    Research output: ThesisDoctoral ThesisCollection of Articles

  5. 2016
  6. Published

    Stochastic gradient estimate variance in contrastive divergence and persistent contrastive divergence

    Berglund, M., 1 Jan 2016, ESANN 2016 - 24th European Symposium on Artificial Neural Networks. p. 521-526 6 p.

    Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

  7. Published

    Scalable gradient-based tuning of continuous regularization hyperparameters

    Luketina, J., Berglund, M., Greff, K. & Raiko, T., 2016, 33rd International Conference on Machine Learning, ICML 2016. Vol. 6. p. 4333-4341 9 p.

    Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

  8. 2015
  9. Published

    Techniques for Learning Binary Stochastic Feedforward Neural Networks

    Raiko, P., Berglund, M., Alain, G. & Dinh, L., 7 May 2015, International Conference on Learning Representations (ICLR 2015). p. 1-10 10 p.

    Research output: Chapter in Book/Report/Conference proceedingConference contributionProfessional

  10. Published

    Measuring the usefulness of hidden units in Boltzmann machines with mutual information

    Berglund, M., Raiko, T. & Cho, K., Apr 2015, In : Neural Networks. 64, p. 12-18 7 p.

    Research output: Contribution to journalArticleScientificpeer-review

  11. Published

    Bidirectional recurrent neural networks as generative models

    Berglund, M., Raiko, T., Honkala, M., Kärkkäinen, L., Vetek, A. & Karhunen, J., 2015, Advances in Neural Information Processing Systems. Neural Information Processing Systems Foundation, Vol. 2015-January. p. 856-864 9 p.

    Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

  12. Published

    Scalable Gradient-Based Tuning of Continuous Regularization Hyperparameters

    Luketina, J., Berglund, M. & Raiko, T., 2015.

    Research output: Working paperProfessional

  13. Published

    Semi-supervised learning with Ladder networks

    Rasmus, A., Valpola, H., Honkala, M., Berglund, M. & Raiko, T., 2015, Advances in Neural Information Processing Systems. Neural Information Processing Systems Foundation, Vol. 2015-January. p. 3546-3554 9 p.

    Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

  14. 2014
  15. Published
  16. Published

    Techniques for Learning Binary Stochastic Feedforward Neural Networks

    Raiko, T., Berglund, M., Alain, G. & Dinh, L., 2014, ArXiv.

    Research output: Working paperProfessional

  17. 2013
  18. Published

    Measuring the Usefulness of Hidden Units in Boltzmann Machines with Mutual Information

    Berglund, M., Raiko, T. & Cho, K., 2013, 20th International Conference, ICONIP 2013, Daegu, Korea, November 3-7, 2013. Proceedings, Part I. p. 482-489

    Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

ID: 75955