Jes Frellsen
Jes Frellsen
Associate Professor of Machine Learning and Signal Processing
Section for Cognitive Systems
Department of Applied Mathematics and Computer Science
Technical University of Denmark (DTU)
Centre for Basic Machine Learning Research in Life Science
Pioneer Centre for Artificial Intelligence
jbayesebayesfbayesrbayes@bayesdbayestbayesubayes.bayesdbayesk

About

I have been an Associate Professor since 2016, and I am currently Associate Professor in Machine Learning and Signal Processing at the Technical University of Denmark (DTU). Before that, I was at the IT University of Copenhagen. Previously I have been a postdoc with Professor Zoubin Ghahramani in the Machine Learning group at the University of Cambridge. Before that, I was a graduate student and postdoc at the Bioinformatics Centre, University of Copenhagen, where I have worked with Professor Anders Krogh and Associate Professor Thomas Hamelryck.

Research interests

I work on statistical machine learning with particular interest in:

  • Generative AI
  • Deep Learning
  • Deep Generative Models
  • Bayesian Modelling and Inference
  • Directional Statistics
  • Markov chain Monte Carlo methods
  • Missing data
  • Application in Bioinformatics

Research group

Postgraduate members:

I am also frequently supervising several BSc and MSc thesis students.

Contact

Phone: +45 4525 3923

Building 321, room 221

Postal address:
Technical University of Denmark
Richard Petersens Plads
Building 321, room 221
2800 Kgs. Lyngby
Denmark

Short biography

Principal academic appointments

Jul 2019 - Associate Professor
Department of Applied Mathematics and Computer Science, Technical University of Denmark
Aug 2016 -
Jun 2019		 Associate Professor
Department of Computer Science, IT University of Copenhagen
May 2013 -
Jun 2016		 Postdoctoral researcher with Zoubin Ghahramani
Department of Engineering, University of Cambridge
Dec 2011 -
Apr 2013		 Postdoctoral researcher with Anders Krogh
Bioinformatics Centre, University of Copenhagen
Mar 2011 -
Aug 2011		 Postdoctoral researcher with Thomas Hamelryck
Bioinformatics Centre, University of Copenhagen

Education

2011 PhD in Bioinformatics
Bioinformatics Centre, University of Copenhagen.
"Probabilistic methods in macromolecular structure prediction"
Supervisor: Thomas Hamelryck, co: Jesper Ferkinghoff-Borg
2007 MSc in Bioinformatics, University of Copenhagen
(I was awarded the highest possible grade for my master thesis)
2005 BSc in maths and CS, University of Copenhagen
2004-2005 EAP exchange student at the University of California, Santa Cruz
(worked in Kevin Karplus's Lab group wither and spring)

Publications

  1. Uppal A, Stensbo-Smidt K, Boomsma W, Frellsen J (2023) Implicit Variational Inference for High-Dimensional Posteriors. Appearing in Advances in Neural Information Processing Systems 37 (NeurIPS 2023). Spotlight presentation. arXiv: 2310.06643
  2. Senetaire HHJ, Garreau D, Frellsen J*, Mattei PA* (2023) Explainability as statistical inference. Proceedings of the 40th International Conference on Machine Learning (ICML 2023), PMLR 202:30584-30612. arXiv: 2212.03131.
    *Equal contribution.
  3. Khomiakov M, Andersen M, Frellsen J (2023) Polygonizer: An auto-regressive building delineator. Machine Learning for Remote Sensing Workshop at ICLR 2023. arXiv: 2304.04048.
  4. Ulmer D, Hardmeier C, Frellsen J (2023) Prior and Posterior Networks: A Survey on Evidential Deep Learning Methods For Uncertainty Estimation. Transactions on Machine Learning Research (TMLR). OpenReview. arXiv: 2110.03051.
  5. Khomiakov M, Mahou AV, Sánchez AR, Frellsen J*, Andersen MR* (2023) Learning to Generate 3D Representations of Building Roofs Using Single-View Aerial Imagery. IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP 2023). doi: 10.1109/ICASSP49357.2023.10095974. arXiv: 2303.11215.
    *Equal contribution.
  6. Zepf K, Petersen E, Frellsen J, Feragen A (2023) That Label's got Style: Handling Label Style Bias for Uncertain Image Segmentation. International Conference on Learning Representations (ICLR 2023). OpenReview.
  7. Bartels S, Stensbo-Smidt K, Moreno-Muñoz P, Boomsma W, Frellsen J, Hauberg S (2023) Adaptive Cholesky Gaussian Processes. Proceedings of The 26th International Conference on Artificial Intelligence and Statistics (AISTATS), PMLR 206:408-452. arXiv: 2202.10769.
  8. Bartels S, Boomsma W, Frellsen J, Garreau D (2023) Kernel-Matrix Determinant Estimates from stopped Cholesky Decomposition. Journal of Machine Learning Research, 24(71):1−57. arXiv: 2107.10587.
  9. Ulmer D, Frellsen J, Hardmeier C (2022) Exploring Predictive Uncertainty and Calibration in NLP: A Study on the Impact of Method & Data Scarcity. Findings of the Association for Computational Linguistics: EMNLP 2022, 2707–2735. ACL Anlanthology, arXiv: 2210.15452.
  10. Ulmer D, Hardmeier C, Frellsen J (2022) deep-significance: Easy and Meaningful Signifcance Testing in the Age of Neural Networks. ML Evaluation Standards Workshop at ICLR 2022. Online version, arXiv: 2204.06815.
  11. Ipsen NB, Mattei P-A, Frellsen J (2022) How to deal with missing data in supervised deep learning? International Conference on Learning Representations (ICLR 2022). OpenReview.
  12. Havtorn JD, Borgholt L, Hauberg S, Frellsen J, Maaløe L (2022) Benchmarking Generative Latent Variable Models for Speech. ICLR Workshop on Deep Generative Models for Highly Structured Data. OpenReview, arXiv: 2202.12707
  13. Bergamin F, Mattei P-A, Havtorn JD, Senetaire H, Schmutz H, Maaløe L, Hauberg S, Frellsen J (2022) Model-agnostic out-of-distribution detection using combined statistical tests. Proceedings of the 25th International Conference on Artificial Intelligence and Statistics (AISTATS), PMLR 151:10753-10776. arXiv: 2203.01097.
  14. Geng C, Wang J, Gao Z, Frellsen J*, Hauberg S* (2021) Bounds all around: training energy-based models with bidirectional bounds. Advances in Neural Information Processing Systems 35 (NeurIPS 2021). Online version, arXiv: 2111.00929.
    *Equal contribution.
  15. Lemaitre P, Andersen MR, Frellsen J (2021) When did the train arrive? A Bayesian approach to enrich timetable information using smart card data. IEEE Open Journal of Intelligent Transportation Systems, 2:160-172. doi: 10.1109/OJITS.2021.3094620.
  16. Havtorn JD, Frellsen J, Hauberg S, Maaløe L (2021) Hierarchical VAEs Know What They Don't Know. Proceedings of the 38th International Conference on Machine Learning (ICML 2021), PMLR 139:4117-4128. Online version, arXiv: 2102.08248.
  17. Ipsen NB, Mattei P-A, Frellsen J (2021) not-MIWAE: Deep Generative Modelling with Missing not at Random Data. International Conference on Learning Representations (ICLR 2021). OpenReview.
  18. McEvoy FJ, Proschowsky HF, Müller A, Moorman L, Bender‐Koch J, Svalastoga EL, Frellsen J, Nielsen DH (2021) Deep transfer learning can be used for the detection of hip joints in pelvis radiographs and the classification of their hip dysplasia status. Veterinary Radiology & Ultrasound, 62:387-393. doi: 10.1111/vru.12968.
  19. Sheiati S, Ranjbar N, Frellsen J, Skare EL, Cepuritis R, Jacobsen S, Spangenberg J (2021) Neural network predictions of the simulated rheological response of cement paste in the FlowCyl. Neural Computing and Applications. doi: 10.1007/s00521-021-05999-4.
  20. Mattei P-A, Frellsen J (2020) Negative Dependence Tightens Variational Bounds. ICML 2020 workshop on Negative Dependence and Submodularity for ML. Online version.
  21. Ipsen NB, Mattei P-A, Frellsen J (2020) How to deal with missing data in supervised deep learning? The first Workshop on the Art of Learning with Missing Values (Artemiss) hosted by the 37th International Conference on Machine Learning (ICML). OpenReview.
  22. Mattei P-A, Frellsen J (2019) MIWAE: Deep Generative Modelling and Imputation of Incomplete Data. Proceedings of the 36th International Conference on Machine Learning (ICML 2019), PMLR 97:4413-4423. Online version, arXiv: 1812.02633.
  23. Wiqvist S, Mattei P-A, Picchini U, Frellsen J (2019) Partially Exchangeable Networks and Architectures for Learning Summary Statistics in Approximate Bayesian Computation. Proceedings of the 36th International Conference on Machine Learning (ICML 2019), PMLR 97:6798-6807. Online version, arXiv: 1901.10230.
  24. Mattei P-A, Frellsen J (2018) Leveraging the Exact Likelihood of Deep Latent Variable Models. Advances in Neural Information Processing Systems 31 (NeurIPS 2018). Online version, arXiv: 1802.04826.
  25. Mattei P-A, Frellsen J (2018) Refit your Encoder when New Data Comes by. Third workshop on Bayesian Deep Learning (at NeurIPS 2018). Online version.
  26. Mattei P-A, Frellsen J (2018) missIWAE: Deep Generative Modelling and Imputation of Incomplete Data. Third workshop on Bayesian Deep Learning (at NeurIPS 2018). Online version.
  27. Mardia KV, Foldager JI, Frellsen J (2018) Directional Statistics in Protein Bioinformatics. In: Ley C, Verdebout T (eds.), Applied Directional Statistics: Modern Methods and Case Studies. CRC Press. doi:10.1201/9781315228570.
  28. Boomsma W, Frellsen J (2017) Spherical convolutions and their application in molecular modelling. Advances in Neural Information Processing Systems 30 (NeurIPS 2017). Spotlight Presentation. Online version.
  29. Brouwer T, Frellsen J, Liò P (2017) Comparative Study of Inference Methods for Bayesian Matrix Factorisation. European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML PKDD) 2017. arXiv: 1707.05147.
  30. Navarro AKW, Frellsen J, Turner RE (2017) The Multivariate Generalised von Mises: Inference and applications. Proceedings of the Thirty-First AAAI Conference on Artificial Intelligence (AAAI-17), 2394-2400. arXiv: 1602.05003.
  31. Brouwer T, Frellsen J, Liò P (2016) Fast Bayesian non-negative matrix factorisation and tri-factorisation. Advances in Approximate Bayesian Inference Workshop at NeurIPS 2016, Barcelona, Spain. arXiv: 1610.08127.
  32. Frellsen J, Winther O, Ghahramani Z, Ferkinghoff-Borg J (2016) Bayesian generalised ensemble Markov chain Monte Carlo. Proceedings of the 19th International Conference on Artificial Intelligence and Statistics (AISTATS), PMLR 51:408-416.
  33. Hamelryck T, Boomsma W, Ferkinghoff-Borg J, Foldager J, Frellsen J, Haslett J, Theobald D (2015) Proteins, physics and probability kinematics: a Bayesian formulation of the protein folding problem. In: Dryden IL, Kent JT (eds.), Geometry Driven Statistics. Wiley. doi:10.1002/9781118866641.ch18.
  34. Boomsma W, Tian P, Frellsen J, Ferkinghoff-Borg J, Hamelryck T, Lindorff-Larsen K, and Vendruscolo M (2014) Equilibrium simulations of proteins using molecular fragment replacement and NMR chemical shifts. PNAS, 111(38):13852-13857. doi:10.1073/pnas.1404948111.
  35. Kerpedjiev P*, Frellsen J*, Lindgreen S, Krogh A (2014) Adaptable probabilistic mapping of short reads using position specific scoring matrices. BMC Bioinformatics 15:100. doi:10.1186/1471-2105-15-100.
    *Joint first author.
  36. Frellsen J*, Menzel P*, Krogh A (2014) Algorithms for mapping high-throughput DNA sequences. In: Brahme A (ed.), Comprehensive Biomedical Physics, Volume 6: Bioinformatics. Elsevier. pp 41-50. doi:10.1016/B978-0-444-53632-7.01103-5.
    *Joint first author.
  37. Frellsen J, Hamelryck T, Ferkinghoff-Borg J (2013) Combining the multicanonical ensemble with generative probabilistic models of local biomolecular structure. Proceedings 59th ISI World Statistics Congress, pp 139-144, 25-30 August 2013, Hong Kong. International Statistical Institute, The Hague, The Netherlands, December 2013. Online version.
  38. Menzel P*, Frellsen J*, Plass M, Rasmussen SH, Krogh A (2013) On the Accuracy of Short Read Mapping. In: Shomron N (ed.), Deep Sequencing Data Analysis, Methods in Molecular Biology. Humana Press. pp. 39-59. doi:10.1007/978-1-62703-514-9_3.
    *Joint first author.
  39. Olsson S, Frellsen J, Boomsma W, Mardia KV, Hamelryck T (2013) Inference of structure ensembles of flexible biomolecules from sparse, averaged data. PLoS ONE, 8(11):e79439. doi:10.1371/journal.pone.0079439.
  40. Valentin JB, Andreetta C, Boomsma W, Bottaro S, Ferkinghoff-Borg J, Frellsen J, Mardia KV, Tian P, Hamelryck T (2013) Formulation of probabilistic models of protein structure in atomic detail using the reference ratio method. Proteins 82:288-299. doi:10.1002/prot.24386.
  41. Boomsma W, Frellsen J, Harder T, Bottaro S, Johansson KE, et al. (2013) PHAISTOS: A framework for Markov chain Monte Carlo simulation and inference of protein structure. Journal of Computational Chemistry 34(19):1697-1705. doi:10.1002/jcc.23292.
  42. Hamelryck T, Haslett J, Mardia K, Kent JT, Valentin J, Frellsen J, Ferkinghoff-Borg J (2013) On the reference ratio method and its application to statistical protein structure prediction. Proceedings of the 32th Leeds Annual Statistical Research Workshop, pp 53-57. Leeds University Press. Online version.
  43. Frellsen J, Mardia KV, Borg M, Ferkinghoff-Borg J, Thomas Hamelryck (2012) Towards a General Probabilistic Model of Protein Structure: The Reference Ratio Method. Hamelryck T et al. (eds.), Bayesian Methods in Structural Bioinformatics, Statistics for Biology and Health. Springer-Verlag. doi:10.1007/978-3-642-27225-7_4.
  44. Mardia KV and Frellsen J (2012) Statistics of Bivariate von Mises Distributions. Hamelryck T et al. (eds.), Bayesian Methods in Structural Bioinformatics, Statistics for Biology and Health. Springer-Verlag. doi:10.1007/978-3-642-27225-7_6 [errata].
  45. Boomsma W, Frellsen J, and Hamelryck T (2012) Probabilistic Models of Local Biomolecular Structure and Their Applications. Hamelryck T et al. (eds.), Bayesian Methods in Structural Bioinformatics, Statistics for Biology and Health. Springer-Verlag. doi:10.1007/978-3-642-27225-7_10.
  46. Olsson S, Boomsma W, Frellsen J, Bottaro S, Harder T, Ferkinghoff-Borg J, Hamelryck T (2011) Generative probabilistic models extend the scope of inferential structure determination. Journal of Magnetic Resonance 213(1):182-186. doi:10.1016/j.jmr.2011.08.039.
  47. Mardia KV, Frellsen J, Borg M, Ferkinghoff-Borg J, Hamelryck T (2011) A statistical view on the reference ratio method. Proceedings of the 30th Leeds Annual Statistical Research Workshop, pp 55-61. Leeds University Press. Online version.
  48. Hamelryck T, Borg M, Paluszewski M, Paulsen, Frellsen J, Andreetta C, Boomsma W, Bottaro S, Ferkinghoff-Borg J (2010) Potentials of Mean Force for Protein Structure Prediction Vindicated, Formalized and Generalized. PLoS ONE, 5(11):e13714. doi:10.1371/journal.pone.0013714.
  49. Harder T, Boomsma W, Paluszewski M, Frellsen J, Johansson KE, Hamelryck T (2010) Beyond rotamers: a generative, probabilistic model of side chains in proteins. BMC Bioinformatics 11:306. doi:10.1186/1471-2105-11-306.
  50. Borg M, Mardia KV, Boomsma W, Frellsen J, Harder T, Stovgaard K, Ferkinghoff-Borg J, Røgen P, Hamelryck T (2009) A probabilistic approach to protein structure prediction: PHAISTOS in CASP9. The 28th Leeds Annual Statistical Research Workshop, pp 65-70. Leeds University Press. Online version.
  51. Frellsen J*, Moltke I*, Thiim M, Mardia KV, Ferkinghoff-Borg J, Hamelryck T (2009) A Probabilistic Model of RNA Conformational Space. PLoS Computational Biology, 5(6):e1000406. doi:10.1371/journal.pcbi.1000406. *Joint first author.
  52. Boomsma W, Borg M, Frellsen J, Harder T, Stovgaard K, Ferkinghoff-Borg J, Krogh A, Mardia KV and Hamelryck, T (2008) PHAISTOS: protein structure prediction using a probabilistic model of local structure. Proceedings of CASP8, pp 82-83. Cagliari, Sardinia, Italy, December 3-7 2008.
  53. Marstrand TT, Frellsen J, Moltke I, Thiim M, Valen E, Retelska D, Krogh A (2008) Asap: A Framework for Over-Representation Statistics for Transcription Factor Binding Sites. PLoS ONE, 3(2):e1623. doi:10.1371/journal.pone.0001623.

Research Group Alumni