Oskar Kviman
Ph.D. student working in the Lagergren Lab at KTH, Stockholm.
email: okviman@kth.se or okviman@gmail.com
Expected to graduate spring 2026 but can graduate autumn 2025
Research interests
Probabilistic machine learning: My recent and current works involve, e.g., variational inference, variational auto-encoders, importance sampling, sequential Monte Carlo methods and Monte Carlo objective functions.
Cancer genomics: Currently I am working on inference of clone distributions based on gene expressions from spatial transcriptomics data.
Phylogenetics: I research variational inference-based phylogenetics and tree sampling methods. Cancer clone tree inference is also of interest.
Conference Publications/Preprints
- Efficient Mixture Learning in Black-Box Variational Inference, ICML 2024 (with Alexandra Hotti, Ricky Molén, Víctor Elvira and Jens Lagergren)
- Indirectly Parameterized Concrete Autoencoders, ICML 2024 (with Alfred Nilsson, Klas Wijk, Sai Bharath Chandra Gutha, Erik Englesson, Alexandra Hotti, Carlo Saccardi, Jens Lagergren, Ricardo Vinuesa and Hossein Azizpour)
- Variational Resampling, AISTATS 2024 (with Nicola Branchini, Víctor Elvira and Jens Lagergren)
- Cooperation in the Latent Space: The Benefits of Adding Mixture Components in Variational Autoencoders, ICML 2023 (with Ricky Molén, Alexandra Hotti, Semih Kurt, Víctor Elvira and Jens Lagergren)
- VaiPhy: a Variational Inference Based Algorithm for Phylogeny, [Oral] NeurIPS 2022 (with Hazal Koptagel, Harald Melin, Negar Safinianaini and Jens Lagergren)
- Multiple Importance Sampling ELBO and Deep Ensembles of Variational Approximations, AISTATS 2022 (with Harald Melin, Hazal Koptagel, Víctor Elvira and Jens Lagergren) [5-min presentation]
- Statistical Distance Based Deterministic Offspring Selection in SMC Methods, arxiv (with Hazal Koptagel, Harald Melin and Jens Lagergren)
- Sequence Disambiguation with Synaptic Traces in Associative Neural Networks, ICANN 2019 (with Ramon H. Martinez, Anders Lansner and Pawel Herman)
Journal Publications
- Improved Variational Bayesian Phylogenetic Inference using Mixtures, TMLR 2024 (with Ricky Molén and Jens Lagergren)
- [Re] Tensor Monte Carlo: Particle Methods for the GPU Era, NeurIPS 2019 Reproducibility Challenge/ReScience-C Journal Publication (with Linus Nilsson and Martin Larsson)
Theses
- KL/TV Reshuffling: Statistical Distance Based Offspring Selection in SMC Methods, Master Thesis 2020 (supervised by Jens Lagergren)
- Applicability of a Translucent Barrier Based Model of Noise, Bachelor Thesis 2018 (with Linus Nilsson; supervised by Saikat Chatterjee)
| Reviewing | Teaching at KTH | Master Thesis Supervision |
|---|---|---|
| AAAI 2025 | Statistical Methods in Applied Computer Science (Lecturer; 2021-) | Xindi Liu (KTH; 2022) |
| Advances in approximate Bayesian inference 2023 | Machine Learning, Advanced Course (TA; 2020-2022) | Ricky Molén (KTH; 2022) |
| NeurIPS 2023 | Deep Learning, Advanced Course (TA; 2021-2022) | |
| AISTATS 2023: rewarded Top reviewer (10%; link) | ||
| NeurIPS 2022 | ||
| ICML 2022 | ||
| AISTATS 2022 | ||
| ICML 2021 |