Publications

Selected publications from the Helsinki Probabilistic Machine Learning Lab appearing in leading machine learning conferences and journals.

2025

[ICLR]         Stochastic variance-reduced Gaussian variational inference on the Bures-Wasserstein manifold
                       Luu H P H, Yu H, Williams B, Hartmann M, Klami A
                       International Conference on Learning Representations 2025
                       ARXIV

[AISTATS]  Noise-Aware Differentially Private Variational Inference
                       Alrawajfeh T, Jälkö J, Honkela A
                       International Conference on Artificial Intelligence and Statistics 2025
                       ARXIV

[AISTATS]  A Bias-Variance Decomposition for Ensembles over Multiple Synthetic Datasets
                       Räisä O, Honkela A
                       International Conference on Artificial Intelligence and Statistics 2025
                       ARXIV

[AISTATS]  Amortized Probabilistic Conditioning for Optimization, Simulation and Inference
                       Chang P E, Loka N, Huang D, Remes U, Kaski S, Acerbi L
                       International Conference on Artificial Intelligence and Statistics 2025
                       ARXIV | CODE

[Journal]  Self-adaptation of ultrasound sensing networks
                       Gharib S, Iablonskyi D, Mustonen J, Korsimaa J, Salminen P, Korkmaz B N, Weber M, Salmi A, Klami A
                       Mechanical Systems and Signal Processing 2025
                       LINK

2024

[NeurIPS]  Non-geodesically-convex optimization in the Wasserstein space
                       Luu H P H, Yu H, Williams B, Mikkola P, Hartmann M, Puolamäki K, Klami A
                       Advances in Neural Information Processing Systems 2024
                       ARXIV

[NeurIPS]  Preferential Normalizing Flows
                       Mikkola P, Acerbi L, Klami A
                       Advances in Neural Information Processing Systems 2024
                       ARXIV | CODE

[NeurIPS]  Amortized Bayesian Experimental Design for Decision-Making
                       Huang D, Guo Y, Acerbi L, Kaski S
                       Advances in Neural Information Processing Systems 2024
                       ARXIV | CODE

[NeurIPS]  Improving robustness to corruptions with multiplicative weight perturbations
                       Trinh T, Heinonen M, Acerbi L, Kaski S
                       Advances in Neural Information Processing Systems 2024
                       ARXIV | CODE

[NeurIPS]  Noise-Aware Differentially Private Regression via Meta-Learning
                       Räisä O, Markou S, Ashman M, Bruinsma W P, Tobaben M, Honkela A, Turner R E
                       Advances in Neural Information Processing Systems 2024
                       ARXIV | CODE

[NeurIPS]  Provable benefits of annealing for estimating normalizing constants: Importance Sampling,
                       Noise-Contrastive Estimation, and beyond
                       Chehab O, Hyvärinen A, Risteski A
                       Advances in Neural Information Processing Systems 2024
                       ARXIV | CODE

[NeurIPS]  LoRANN: Low-Rank Matrix Factorization for Approximate Nearest Neighbor Search
                       Jääsaari E, Hyvönen V, Roos T
                       Advances in Neural Information Processing Systems 2024
                       ARXIV | CODE

[UAI]           Faster Perfect Sampling of Bayesian Network Structures
                       Harviainen J, Koivisto M
                       Conference on Uncertainty in Artificial Intelligence 2024
                       LINK

[ICML]         Causal Representation Learning Made Identifiable by Grouping of Observational Variables
                       Morioka H, Hyvarinen A
                       International Conference on Machine Learning 2024
                       ARXIV | CODE

[ICML]         Estimating the Permanent by Nesting Importance Sampling
                       Harviainen J, Koivisto M
                       International Conference on Machine Learning 2024
                       LINK

[ICLR]         Input-gradient space particle inference for neural network ensembles
                       Trinh T, Heinonen M, Acerbi L, Kaski S
                       International Conference on Learning Representations 2024
                       ARXIV | CODE

[AISTATS]  Riemannian Laplace Approximation with the Fisher Metric
                       Yu H, Hartmann M, Williams B, Girolami M, Klami A
                       International Conference on Artificial Intelligence and Statistics 2024
                       ARXIV | CODE

[AISTATS]  Identifiable Feature Learning for Spatial Data with Nonlinear ICA
                       Hälvä H, So J, Turner R E, Hyvärinen A
                       International Conference on Artificial Intelligence and Statistics 2024
                       ARXIV

[AISTATS]  Error bounds for any regression model using Gaussian processes with gradient information
                       Savvides R, Luu H P H, Puolamäki K
                       International Conference on Artificial Intelligence and Statistics 2024
                       LINK | CODE

[JMLR]         A Multilabel Classification Framework for Approximate Nearest Neighbor Search
                       Hyvönen V, Jääsaari E, Roos T
                       Journal of Machine Learning Research 2024
                       LINK

[JAIR]         Approximate counting of linear extensions in practice
                       Talvitie T, Koivisto M
                       Journal of Artificial Intelligence Research 2024
                       LINK | CODE

[Journal]  Prior knowledge elicitation: The past, present, and future
                       Mikkola P, Martin O A, Chandramouli S, Hartmann M, Abril Pla O, Thomas O, Pesonen H, Corander J,
                       Vehtari A, Kaski S, Bürkner P, Klami A
                       Bayesian Analysis 2024
                       LINK

2023

[NeurIPS]  Practical equivariances via relational conditional neural processes
                       Huang D, Haussmann M, Remes U, John S, Clarté G, Luck K, Kaski S, Acerbi L
                       Advances in Neural Information Processing Systems 2023
                       ARXIV | CODE

[NeurIPS]  Learning robust statistics for simulation-based inference under model misspecification
                       Huang D, Bharti A, Souza A, Acerbi L, Kaski S
                       Advances in Neural Information Processing Systems 2023
                       ARXIV | CODE

[UAI]           Revisiting Bayesian network learning with small vertex cover
                       Harviainen J, Koivisto M
                       Uncertainty in Artificial Intelligence 2023
                       LINK

[UAI]           On inference and learning with probabilistic generating circuits
                       Harviainen J, Ramaswamy V P, Koivisto M
                       Uncertainty in Artificial Intelligence 2023
                       LINK

[ICML]         Estimating the contamination factor’s distribution in unsupervised anomaly detection
                       Perini L, Bürkner P, Klami A
                       International Conference on Machine Learning 2023
                       ARXIV | CODE

[AISTATS]  Noise-aware statistical inference with differentially private synthetic data
                       Räisä O, Jälkö J, Kaski S, Honkela A
                       International Conference on Artificial Intelligence and Statistics 2023
                       ARXIV | CODE

[AISTATS]  Connectivity-contrastive learning: Combining causal discovery and representation learning for
                       multimodal data
                       Morioka H, Hyvärinen A
                       International Conference on Artificial Intelligence and Statistics 2023
                       LINK | CODE

[AAAI]         A faster practical approximation scheme for the permanent
                       Harviainen J, Koivisto M
                       Proceedings of the AAAI Conference on Artificial Intelligence 2023
                       LINK

[JMLR]         Prior specification for Bayesian matrix factorization via prior predictive matching
                       da Silva E d S, Kuśmierczyk T, Hartmann M, Klami A
                       Journal of Machine Learning Research 2023
                       ARXIV | CODE

[Journal]  Exploring uplift modeling with high class imbalance
                       Nyberg O, Klami A
                       Data Mining and Knowledge Discovery 2023
                       LINK | CODE

[Journal]  Unsupervised representation learning of spontaneous MEG data with nonlinear ICA
                       Zhu Y, Parviainen T, Heinilä E, Parkkonen L, Hyvärinen A
                       NeuroImage 2023
                       LINK

[Journal]  SLISEMAP: Supervised dimensionality reduction through local explanations
                       Björklund A, Mäkelä J, Puolamäki K
                       Machine Learning 2023
                       ARXIV | CODE

2022

[NeurIPS]  Trustworthy Monte Carlo
                       Harviainen J, Koivisto M, Kaski P
                       Advances in Neural Information Processing Systems 2022
                       LINK

[NeurIPS]  A multilabel classification framework for approximate nearest neighbor search
                       Hyvönen V, Jääsaari E, Roos T
                       Advances in Neural Information Processing Systems 2022
                       ARXIV | CODE

[UAI]           The optimal noise in noise-contrastive learning is not what you think
                       Chehab O, Gramfort A, Hyvärinen A
                       Uncertainty in Artificial Intelligence 2022
                       ARXIV | CODE

[UAI]           Binary independent component analysis: a non-stationarity-based approach
                       Hyttinen A, Pacela V B, Hyvärinen A
                       Uncertainty in Artificial Intelligence 2022
                       ARXIV | CODE

[ICML]         Tackling covariate shift with node-based Bayesian neural networks
                       Trinh T Q, Heinonen M, Acerbi L, Kaski S
                       International Conference on Machine Learning 2022
                       ARXIV | CODE

[AISTATS]  Parallel MCMC without embarrassing failures
                       De Souza D A, Mesquita D, Kaski S, Acerbi L
                       International Conference on Artificial Intelligence and Statistics 2022
                       ARXIV | CODE

[AISTATS]  Lagrangian manifold Monte Carlo on Monge patches
                       Hartmann M, Girolami M, Klami A
                       International Conference on Artificial Intelligence and Statistics 2022
                       ARXIV | CODE

[Journal]  Strong pathogen competition in neonatal gut colonisation
                       Mäklin T, Thorpe H A, Pöntinen A K, Gladstone R A, Shao Y, Pesonen M, McNally A, Johnsen P J,
                       Samuelsen Ø, Lawley T D, Honkela A, Corander J
                       Nature Communications 2022
                       LINK

[Journal]  Dynamics of retinotopic spatial attention revealed by multifocal MEG
                       Kurki I, Hyvärinen A, Henriksson L
                       NeuroImage 2022
                       LINK

[Journal]  Visual data exploration as a statistical testing procedure: Within-view and between-view multiple
                       comparisons
                       Savvides R, Henelius A, Oikarinen E, Puolamäki K
                       IEEE Transactions on Visualization and Computer Graphics 2022
                       LINK | CODE

[Journal]  Robust regression via error tolerance
                       Björklund A, Henelius A, Oikarinen E, Kallonen K, Puolamäki K
                       Data Mining and Knowledge Discovery 2022
                       LINK

2021

[NeurIPS]  Shared independent component analysis for multi-subject neuroimaging
                       Richard H, Ablin P, Thirion B, Gramfort A, Hyvärinen A
                       Advances in Neural Information Processing Systems 2021
                       ARXIV | CODE

[NeurIPS]  Disentangling identifiable features from noisy data with structured nonlinear ICA
                       Hälvä H, Le Corff S, Lehéricy L, So J, Zhu Y, Gassiat E, Hyvärinen A
                       Advances in Neural Information Processing Systems 2021
                       ARXIV | CODE

[NeurIPS]  Approximating the permanent with deep rejection sampling
                       Harviainen J, Röyskö A, Koivisto M
                       Advances in Neural Information Processing Systems 2021
                       ARXIV | CODE

[ICML]         Differentially private Bayesian inference for generalized linear models
                       Kulkarni T, Jälkö J, Koskela A, Kaski S, Honkela A
                       International Conference on Machine Learning 2021
                       ARXIV

[AISTATS]  Tight differential privacy for discrete-valued mechanisms and for the subsampled Gaussian mechanism
                       using fft
                       Koskela A, Jälkö J, Prediger L, Honkela A
                       International Conference on Artificial Intelligence and Statistics 2021
                       ARXIV

[AISTATS]  Causal autoregressive flows
                       Khemakhem I, Monti R, Leech R, Hyvärinen A
                       International Conference on Artificial Intelligence and Statistics 2021
                       ARXIV | CODE

[AISTATS]  Independent innovation analysis for nonlinear vector autoregressive process
                       Morioka H, Hälvä H, Hyvärinen A
                       International Conference on Artificial Intelligence and Statistics 2021
                       ARXIV | CODE

[JMLR]         Information criteria for non-normalized models
                       Matsuda T, Uehara M, Hyvärinen A
                       Journal of Machine Learning Research 2021
                       ARXIV

[JMLR]         Guided visual exploration of relations in data sets
                       Puolamäki K, Oikarinen E, Henelius A
                       Journal of Machine Learning Research 2021
                       LINK

[Journal]  Direction matters: On influence-preserving graph summarization and max-cut principle for directed
                       graphs
                       Xu W, Niu G, Hyvärinen A, Sugiyama M
                       Neural Computation 2021
                       ARXIV

[Journal]  Machine-learning models to replicate large-eddy simulations of air pollutant concentrations along
                       boulevard-type streets
                       Lange M, Suominen H, Kurppa M, Järvi L, Oikarinen E, Savvides R, Puolamäki K
                       Geoscientific Model Development 2021
                       LINK

[Journal]  Detecting virtual concept drift of regressors without ground truth values
                       Oikarinen E, Tiittanen H, Henelius A, Puolamäki K
                       Data Mining and Knowledge Discovery 2021
                       LINK

[Journal]  Gradient-based training and pruning of radial basis function networks with an application in
                       materials physics
                       Määttä J, Bazaliy V, Kimari J, Djurabekova F, Nordlund K, Roos T
                       Neural Networks 2021
                       LINK | CODE

2020

[NeurIPS]  Dynamic allocation of limited memory resources in reinforcement learning
                       Patel N, Acerbi L, Pouget A
                       Advances in Neural Information Processing Systems 2020
                       ARXIV | CODE

[NeurIPS]  Variational Bayesian Monte Carlo with noisy likelihoods
                       Acerbi L
                       Advances in Neural Information Processing Systems 2020
                       ARXIV | CODE

[NeurIPS]  Relative gradient optimization of the jacobian term in unsupervised deep learning
                       Gresele L, Fissore G, Javaloy A, Schölkopf B, Hyvärinen A
                       Advances in Neural Information Processing Systems 2020
                       ARXIV | CODE

[NeurIPS]  Modeling shared responses in neuroimaging studies through multiview ica
                       Richard H, Gresele L, Hyvärinen A, Thirion B, Gramfort A, Ablin P
                       Advances in Neural Information Processing Systems 2020
                       ARXIV | CODE

[NeurIPS]  Ice-beem: Identifiable conditional energy-based deep models based on nonlinear ica
                       Khemakhem I, Monti R, Kingma D, Hyvärinen A
                       Advances in Neural Information Processing Systems 2020
                       ARXIV | CODE

[NeurIPS]  Towards scalable bayesian learning of causal dags
                       Viinikka J, Hyttinen A, Pensar J, Koivisto M
                       Advances in Neural Information Processing Systems 2020
                       ARXIV | CODE

[UAI]           Sensor placement for spatial Gaussian processes with integral observations
                       Longi K, Rajani C, Sillanpää T, Mäkinen J, Rauhala T, Salmi A, Hæggström E, Klami A
                       Conference on Uncertainty in Artificial Intelligence 2020
                       LINK | CODE

[UAI]           Flexible prior elicitation via the prior predictive distribution
                       Hartmann M, Agiashvili G, Bürkner P, Klami A
                       Conference on Uncertainty in Artificial Intelligence 2020
                       ARXIV

[UAI]           Hidden markov nonlinear ica: Unsupervised learning from nonstationary time series
                       Hälvä H, Hyvärinen A
                       Conference on Uncertainty in Artificial Intelligence 2020
                       ARXIV | CODE

[UAI]           Robust contrastive learning and nonlinear ICA in the presence of outliers
                       Sasaki H, Takenouchi T, Monti R, Hyvärinen A
                       Conference on Uncertainty in Artificial Intelligence 2020
                       ARXIV

[UAI]           Causal discovery with general non-linear relationships using non-linear ica
                       Monti R P, Zhang K, Hyvärinen A
                       Uncertainty in Artificial Intelligence 2020
                       ARXIV

[UAI]           Layering-mcmc for structure learning in bayesian networks
                       Viinikka J, Koivisto M
                       Conference on Uncertainty in Artificial Intelligence 2020
                       LINK | CODE

[UAI]           Exact sampling of directed acyclic graphs from modular distributions
                       Talvitie T, Vuoksenmaa A, Koivisto M
                       Uncertainty in Artificial Intelligence 2020
                       LINK | CODE

[AISTATS]  Learning rate adaptation for differentially private learning
                       Koskela A, Honkela A
                       International Conference on Artificial Intelligence and Statistics 2020
                       ARXIV | CODE

[AISTATS]  Computing tight differential privacy guarantees using fft
                       Koskela A, Jälkö J, Honkela A
                       International Conference on Artificial Intelligence and Statistics 2020
                       ARXIV

[AISTATS]  Variational autoencoders and nonlinear ica: A unifying framework
                       Khemakhem I, Kingma D, Monti R, Hyvärinen A
                       International Conference on Artificial Intelligence and Statistics 2020
                       ARXIV | CODE

[AAAI]         Correcting predictions for approximate Bayesian inference
                       Kuśmierczyk T, Sakaya J, Klami A
                       Proceedings of the AAAI Conference on Artificial Intelligence 2020
                       ARXIV

[AAAI]         Error-correcting and verifiable parallel inference in graphical models
                       Karimi N, Kaski P, Koivisto M
                       Proceedings of the AAAI Conference on Artificial Intelligence 2020
                       LINK | CODE

[AAAI]         A Bayesian approach for estimating causal effects from observational data
                       Pensar J, Talvitie T, Hyttinen A, Koivisto M
                       Proceedings of the AAAI Conference on Artificial Intelligence 2020
                       LINK

[Journal]  Unbiased and efficient log-likelihood estimation with inverse binomial sampling
                       van Opheusden B, Acerbi L, Ma W J
                       PLoS Computational Biology 2020
                       LINK | CODE

[Journal]  The role of sensory uncertainty in simple contour integration
                       Zhou Y, Acerbi L, Ma W J
                       PLoS Computational Biology 2020
                       LINK

[Journal]  Multiscale cloud detection in remote sensing images using a dual convolutional neural network
                       Luotamo M, Metsämäki S, Klami A
                       IEEE Transactions on Geoscience and Remote Sensing 2020
                       LINK

[Journal]  Brain activity reflects the predictability of word sequences in listened continuous speech
                       Koskinen M, Kurimo M, Gross J, Hyvärinen A, Hari R
                       NeuroImage 2020
                       LINK

[Journal]  Nonlinear ICA of fMRI reveals primitive temporal structures linked to rest, task, and behavioral
                       traits
                       Morioka H, Calhoun V, Hyvärinen A
                       NeuroImage 2020
                       LINK

[Journal]  Interactive visual data exploration with subjective feedback: an information-theoretic approach
                       Puolamäki K, Oikarinen E, Kang B, Lijffijt J, De Bie T
                       Data Mining and Knowledge Discovery 2020
                       ARXIV | CODE

[Journal]  Nonlinear dimensionality reduction for clustering
                       Sotiris Tasoulis, Nicos G. Pavlidis, Teemu Roos
                       Pattern Recognition 2020
                       LINK