Avatar
hlynur.hlynsson@ini.rub.de (+49) 234-32-27988

Hlynur Davíð Hlynsson, M.Sc.

Theory of Neural Systems

Theory of Neural Systems

hlynur.hlynsson@ini.rub.de

(+49) 234-32-27988

I'm a PhD student at the Institute Für Neuroinformatik and a member of the Theory of Neural Systems research group.

My research interests are: representation learning, deep learning, self-supervised learning, deep reinforcement learning.

  • 2011 - 2015

    BSc Computer Science

    The University of Iceland, First Grade

  • 2011 - 2015

    BSc Mathematics

    The University of Iceland, First Grade

  • 2015 - 2017

    MSc Machine Learning

    KTH Royal Institute of Technology

    2022

  • Latent Representation Prediction Networks
    Hlynsson, H. D., Schüler, M., Schiewer, R., Glasmachers, T., & Wiskott, L.
    International Journal of Pattern Recognition and Artificial Intelligence, 36(01), 2251002
  • Reduction of Variance-related Error through Ensembling: Deep Double Descent and Out-of-Distribution Generalization
    Rath-Manakidis, P., Hlynsson, H. D., & Wiskott, L.
    In Proceedings of the 11th International Conference on Pattern Recognition Applications and Methods - ICPRAM, (pp. 31–40) SciTePress
    pdf

    • 2021

  • Reward prediction for representation learning and reward shaping
    Hlynsson, H. D., & Wiskott, L.
    arXiv

    • 2020

  • Latent Representation Prediction Networks
    Hlynsson, H. D., Schüler, M., Schiewer, R., Glasmachers, T., & Wiskott, L.
    arXiv preprint arXiv:2009.09439
    Latent Representation Prediction Networks

    • 2019

  • Measuring the Data Efficiency of Deep Learning Methods
    Hlynsson, H., Escalante-B., A., & Wiskott, L.
    In Proceedings of the 8th International Conference on Pattern Recognition Applications and Methods SCITEPRESS - Science and Technology Publications
    Measuring the Data Efficiency of Deep Learning Methods Conference poster
  • Learning Gradient-Based ICA by Neurally Estimating Mutual Information
    Hlynsson, H. D. ′ið, & Wiskott, L.
    In C. Benzmüller & Stuckenschmidt, H. (Eds.), KI 2019: Advances in Artificial Intelligence (pp. 182–187) Cham: Springer International Publishing
    link link
  • Learning gradient-based ICA by neurally estimating mutual information
    Hlynsson, H. D., & Wiskott, L.
    arXiv, arXiv–1904.09858
    pdf link
  • Measuring the Data Efficiency of Deep Learning Methods
    Hlynsson, H. D., Wiskott, L., & others,
    arXiv, arXiv–1907
    pdf
  • Gradient-based Training of Slow Feature Analysis by Differentiable Approximate Whitening
    Schüler, M., Hlynsson, H. D. ′ið, & Wiskott, L.
    In W. S. Lee & Suzuki, T. (Eds.), Proceedings of The Eleventh Asian Conference on Machine Learning (Vol. 101, pp. 316–331) Nagoya, Japan: PMLR
    link

    • 2018

  • Gradient-based Training of Slow Feature Analysis by Differentiable Approximate Whitening
    Schüler, M., Hlynsson, H. D., & Wiskott, L.
    CoRR e-print arXiv:1808.08833

Summer Term 2020

Lab coursesIntroduction to Python

Summer Term 2019

Lab coursesIntroduction to Python

Summer Term 2018

Lab coursesIntroduction to Python

    2021

  • Interaction of Ensembling and Double Descent in Deep Neural Networks
    Rath-Manakidis, P.
    Master’s thesis, Cognitive Science, Ruhr University Bochum, Germany

A brief introduction to Slow Feature Analysis

One of the main research topics of the TNS group is called Slow Feature Analysis. Slow feature analysis (SFA) is an unsupervised learning method to extract the slowest or smoothest underlying functions or features from a time series. This can be used for dimensionality reduction, regression and classification. In this post we will provide a code example where SFA is applied, to help motivate the method. Then we will go into more detail about the math behind the method and finally provide links to other good resources on the material.
2020

Exploration of Deep Double Descent through Ensemble Formation

We empirically decompose the generalization loss of deep neural networks into bias and variance components on an image classification task by constructing ensembles using geometric-mean averaging of the sub-net outputs and we isolate double descent at the variance component of the loss. Our results show that small models afford ensembles that outperform single large models while requiring considerably fewer parameters and computational steps. We also find that deep double descent that depends on the existence of label noise can be mitigated by using ensembles of models subject to identical label noise almost as thoroughly as by ensembles of networks each trained subject to i.i.d. noise.
2019

Predicting Latent Space Representations for Planning

Pairing the EfferenceNet with a good but generic feature map allows us to perform an accurate search in the latent space of manipulating unseen objects. This remarkably simple method, inspired by the neurology of the cerebellum, reveals a promising line of future work. We validate our method by on a viewpoint-matching task derived from the NORB data set.
2018

Learning gradient-based ICA by neurally estimating mutual information

Several methods of estimating the mutual information of random variables have been developed in recent years. They can prove valuable for novel approaches to learning statistically independent features. In this paper, we use one of these methods, a mutual information neural estimation (MINE) network, to present a proof-of-concept of how a neural network can perform linear ICA. We minimize the mutual information,as estimated by a MINE network, between the output units of a differentiable encoder network. This is done by simple alternate optimization of the two networks. The method is shown to get a qualitatively equal solution to FastICA on blind-source-separation of noisy sources.
2018

Measuring the Data Efficiency of Deep Learning Methods

We propose a new experimental protocol and use it to benchmark the data efficiency — performance as a function of training set size — of two deep learning algorithms, convolutional neural networks (CNNs) and hierarchical information-preserving graph-based slow feature analysis (HiGSFA), for tasks in classification and transfer learning scenarios.

About the INI

The Institut für Neuroinformatik (INI) is a central research unit of the Ruhr-Universität Bochum. We aim to understand the fundamental principles through which organisms generate behavior and cognition while linked to their environments through sensory systems and while acting in those environments through effector systems. Inspired by our insights into such natural cognitive systems, we seek new solutions to problems of information processing in artificial cognitive systems. We draw from a variety of disciplines that include experimental approaches from psychology and neurophysiology as well as theoretical approaches from physics, mathematics, electrical engineering and applied computer science, in particular machine learning, artificial intelligence, and computer vision.

Contact

Universitätsstr. 150, Building NB, Room 3/32
D-44801 Bochum, Germany

Tel: (+49) 234 32-28967
Fax: (+49) 234 32-14210

sekretariat@ini.rub.de
www.ini.rub.de