Next week we have a guest from Linköping.
Fredrik Lindsten, Associate Professor at the Division of Statistics and Machine Learning, Department of Computer and Information Science, Linköping University, will talk about his recent ICML paper.
Gaussian Markov random fields (GMRFs) are probabilistic graphical models widely used in spatial statistics and related fields to model dependencies over spatial structures. In this talk I present a formal connection between GMRFs and convolutional neural networks (CNNs). Common GMRFs are shown to be special cases of a generative model where the inverse mapping from data to latent variables is given by a 1-layer linear CNN. This connection allows us to generalize GMRFs to multi-layer CNN architectures, effectively increasing the order of the corresponding GMRF in a way which has favorable computational scaling. Furthermore, well-established tools, such as autodiff and variational inference, can be used for simple and efficient inference and learning of the deep GMRF. I demonstrate the flexibility of the proposed model and show that it outperforms the state-of-the-art on a dataset of satellite temperatures, in terms of prediction and predictive uncertainty.
The paper is available here.
Short bio of the speaker:
Fredrik Lindsten is Associate Professor at the Division of Statistics and Machine Learning, Department of Computer and Information Science, Linköping University, Sweden. He received his MSc degree in Applied Physics and Electrical Engineering in 2008 and a PhD in Automatic Control in 2013, both from Linköping University. In 2014 and 2015 he was a Postdoctoral Research Associate at the Department of Engineering, the University of Cambridge, UK. During spring 2012 he was a Visiting Student Researcher at the Statistical Artificial Intelligence Lab at the University of California, Berkeley, USA and during spring 2015 he was a Visiting Scholar at the Department of Statistics, the University of Oxford, UK. He has received the Ingvar Carlsson Award by the Swedish Foundation for Strategic Research, and the Benzelius Award by the Royal Society of Sciences in Uppsala. Lindsten's main research interests are in statistical machine learning and computational statistics.
You may join the meeting via the following link:
Chalmers machine learning seminars are organised by the division of Data Science and AI and open to the public with speakers from both academia and industry. Feel free to reach out to us if you have something that you think would be interesting to present.
Emil Carlsson, caremil(at)chalmers.se
Arman Rahbar, armanr(at)chalmers.se
16 November, 2020, 14:00
16 November, 2020, 15:00