Computing Tight Differential Privacy Guarantees Using FFT A Koskela, J Jälkö, A Honkela
International Conference on Artificial Intelligence and Statistics, 2560-2569, 2020
106 2020 Ethylene glycol revisited: Molecular dynamics simulations and visualization of the liquid and its hydrogen-bond network A Kaiser, O Ismailova, A Koskela, SE Huber, M Ritter, B Cosenza, ...
Journal of Molecular Liquids 189, 20-29, 2014
71 2014 Tight differential privacy for discrete-valued mechanisms and for the subsampled gaussian mechanism using FFT A Koskela, J Jälkö, L Prediger, A Honkela
International Conference on Artificial Intelligence and Statistics, 3358-3366, 2021
62 2021 Learning rate adaptation for differentially private learning A Koskela, A Honkela
International Conference on Artificial Intelligence and Statistics, 2465-2475, 2020
49 * 2020 Differentially private cross-silo federated learning MA Heikkilä, A Koskela, K Shimizu, S Kaski, A Honkela
arXiv preprint arXiv:2007.05553, 2020
36 2020 Differentially private Bayesian inference for generalized linear models T Kulkarni, J Jälkö, A Koskela, S Kaski, A Honkela
International Conference on Machine Learning, 5838-5849, 2021
35 2021 Exponential Taylor methods: Analysis and implementation A Koskela, A Ostermann
Computers & Mathematics with Applications 65 (3), 487-499, 2013
24 2013 Splitting methods for time integration of trajectories in combined electric and magnetic fields C Knapp, A Kendl, A Koskela, A Ostermann
Physical Review E 92 (6), 063310, 2015
22 2015 Numerical Accounting in the Shuffle Model of Differential Privacy A Koskela, M Heikkilä, A Honkela
Transactions on Machine Learning Research, 2023
21 * 2023 Analysis of Krylov Subspace Approximation to Large Scale Differential Riccati Equations A Koskela, H Mena
Electronic Transactions on Numerical Analysis 52, 431--454, 2020
21 * 2020 Computing low-rank approximations of the Fréchet derivative of a matrix function using Krylov subspace methods P Kandolf, A Koskela, SD Relton, M Schweitzer
Numerical Linear Algebra with Applications, e2401, 2021
20 2021 Approximating the matrix exponential of an advection-diffusion operator using the incomplete orthogonalization method A Koskela
Numerical Mathematics and Advanced Applications-ENUMATH 2013: Proceedings of …, 2014
16 2014 Computing differential privacy guarantees for heterogeneous compositions using FFT A Koskela, A Honkela
arXiv preprint arXiv:2102.12412, 2021
15 2021 Individual Privacy Accounting with Gaussian Differential Privacy A Koskela, M Tobaben, A Honkela
International Conference on Learning Representations, 2023
13 2023 Disguised and new quasi-Newton methods for nonlinear eigenvalue problems E Jarlebring, A Koskela, G Mele
Numerical Algorithms 79, 311-335, 2018
11 2018 Differentially private hamiltonian monte carlo O Räisä, A Koskela, A Honkela
arXiv preprint arXiv:2106.09376, 2021
7 2021 Krylov integrators for Hamiltonian systems T Eirola, A Koskela
BIT Numerical Mathematics 59, 57-76, 2019
7 2019 Practical differentially private hyperparameter tuning with subsampling A Koskela, T Kulkarni
Thirty-seventh Conference on Neural Information Processing Systems, 2023
6 2023 The infinite Arnoldi exponential integrator for linear inhomogeneous ODEs A Koskela, E Jarlebring
arXiv preprint arXiv:1502.01613, 2015
5 2015 A Moment-Matching Arnoldi Iteration for Linear Combinations of Functions A Koskela, A Ostermann
SIAM Journal on Matrix Analysis and Applications 35 (4), 1344-1363, 2014
4 2014