Publications
Articles
Kernel-based linear system identification using augmented Krylov subspaces
Fabio Matti, Martin S. Andersen, Tianshi Chen, Daniel Kressner. 2026. [arXiv, GitHub]
Attention Mechanisms Through the Lens of Numerical Methods: Approximation Methods and Alternative Formulations
Michel F. Serret, Alice Cortinovis, Yijun Dong, Diana Halikias, Anna Ma, Fabio Matti, Deanna Needell, Katherine J. Pearce, Elizaveta Rebrova, Disha Shur, Rudi Smith, Hai-Xiao Wang, Laura Grigori. 2026. [arXiv]
Stochastic trace estimation for parameter-dependent matrices applied to spectral density approximation
Fabio Matti, Haoze He, Daniel Kressner, Hei Yin Lam. 2026. SIAM Journal on Scientific Computing. [SIAM J. Sci. Comp., arXiv, GitHub]
A Click Ahead: Real-Time Forecasting of Keyboard and Mouse Actions using RNNs and Computer Vision
Fabio Matti, Pierre Dillenbourg, Ludovico Novelli. 2023. [arXiv]
Talks
Stochastic trace estimation for parameter-dependent matrices applied to spectral density approximation
ILAS 2026, Blacksburg VA, USA. [pdf]
Linear system identification using augmented Krylov subspace methods
SWICCOMAS Annual Meeting 2026, Bern, Switzerland. [pdf]
Krylov-augmented finite impulse response estimation
GAMM ANLA 2025, Bologna, Italy. [pdf]
Posters
Linear system identification using augmented Krylov subspace methods
FoCM 2026, Vienna, Austria. [pdf]
A medley of fun techniques for solving interesting problems in RandNLA
RandNLA Workshop 2026, Lausanne, Switzerland. [pdf]
Randomized spectral density estimation applied to neural network optimization
LoRAINNe’24: LOw-Rank Approximations and their Interactions with Neural NEtworks. 2024. [pdf]
Randomized trace estimation for parameter-dependent matrices applied to spectral density approximation
Swiss Numerical Analysis Day. 2024. [pdf]
Theses
Randomized estimation of spectral densities
Master’s thesis. 2024. [thesis, slides, poster, code]
Gradient boosting with explainable wavelength selection for learned spectral decoloring in optoacoustic imaging
Bachelor’s thesis. 2021. [thesis, code]