arXiv subjects

Berent Ånund Strømnes Lunde

Publications and source records attributed to Berent Ånund Strømnes Lunde.

5 recordsLinked to original sources

arXiv2026

Learning the Structure of Triangular Transport Maps

Triangular transport maps provide a flexible approach to sampling-based probabilistic modeling, including density estimation, generative modeling, and Bayesian inference. They transform an unknown target distribution into a simpler reference through a monotone triangular map. The map structure is defined by a variable ordering and sparsity pattern, which together encode a directed acyclic graph. Map quality can depend strongly on this structure, yet finding a good structure is computationally expensive because each candidate generally requires fitting a different map. A central challenge is therefore to learn density and structure jointly, while keeping computation manageable as dimension grows. We introduce Self-Structuring Transport Maps (SSTM), which learn the map, ordering, and sparsity jointly. We use SoftSort to learn the variable ordering and $L_0$ gates to learn the sparsity, while preserving a triangular structure. To keep the map scalable, we use a monotone BatchEnsemble that shares one weight matrix across all map components through rank-one adapters. Across synthetic and real data, jointly learning the structure and map gives better density estimates than estimating the structure first. When the structure is identifiable from the density, SSTM matches the density performance of a map fitted with the true structure and outperforms autoregressive flows. On large datasets, SSTM is competitive with autoregressive flows.

cs.LG↗

arXiv2026

An Ensemble Information Filter: Learning and Encoding Sparse Structures in Ensemble Kalman Methods

Ensemble-based Data Assimilation faces significant challenges in high-dimensional systems due to spurious correlations and ensemble collapse. We propose the Ensemble Information Filter, which addresses these issues by encoding locality directly in the precision matrix through sparse Markov structure. This approach is motivated by connections to SPDE-based models, where local operators induce sparse precision representations and provide a principled form of regularisation. The framework further incorporates sparse regression for learning the observation operator, enabling stable inference in regimes where $p \gg n$. By enforcing structure prior to estimation, EnIF avoids the need for heuristic localisation and improves both statistical and computational scalability. The methodology is demonstrated on filtering, smoothing, and parameter estimation problems, where it adapts to the underlying dependence structure and yields stable and efficient inference in high-dimensional systems.

stat.ME↗

arXiv2022

GraphSPME: Markov Precision Matrix Estimation and Asymptotic Stein-Type Shrinkage

GraphSPME is an open source Python, R and C++ header-only package implement-ing non-parametric sparse precision matrix estimation along with asymptotic Stein-type shrinkage estimation of the covariance matrix. The user defines a potential neighbourhood structure and provides data that potentially are p >> n. This paper introduces a novel approach for finding the optimal order (that data allows to estimate) of a potential Markov property. The algorithm is implemented in the package, alleviating the problem of users making Markov assumptions and implementing corresponding complex higher-order neighbourhood structures. Estimation is made accurate and stable by simultaneously utilising both Markov properties and Stein-type shrinkage. Asymptotic results on Stein-type shrinkage ensure that non-singular well conditioned matrices are obtained in an automatic manner. Final symmetry conversion creates symmetric positive definite estimates. Furthermore, the estimation routine is made efficient and scalable to very high-dimensional problems (~10^7) by utilising the sparse nature of the precision matrix under Markov assumptions. Implementation wise, the sparsity is exploited by employing the sparsity possibilities made available by the Eigen C++ linear-algebra library. The package and examples are available at https://github.com/equinor/GraphSPME

stat.CO↗

arXiv2020

agtboost: Adaptive and Automatic Gradient Tree Boosting Computations

agtboost is an R package implementing fast gradient tree boosting computations in a manner similar to other established frameworks such as xgboost and LightGBM, but with significant decreases in computation time and required mathematical and technical knowledge. The package automatically takes care of split/no-split decisions and selects the number of trees in the gradient tree boosting ensemble, i.e., agtboost adapts the complexity of the ensemble automatically to the information in the data. All of this is done during a single training run, which is made possible by utilizing developments in information theory for tree algorithms {\tt arXiv:2008.05926v1 [stat.ME]}. agtboost also comes with a feature importance function that eliminates the common practice of inserting noise features. Further, a useful model validation function performs the Kolmogorov-Smirnov test on the learned distribution.

stat.ML↗

arXiv2020

An information criterion for automatic gradient tree boosting

An information theoretic approach to learning the complexity of classification and regression trees and the number of trees in gradient tree boosting is proposed. The optimism (test loss minus training loss) of the greedy leaf splitting procedure is shown to be the maximum of a Cox-Ingersoll-Ross process, from which a generalization-error based information criterion is formed. The proposed procedure allows fast local model selection without cross validation based hyper parameter tuning, and hence efficient and automatic comparison among the large number of models performed during each boosting iteration. Relative to xgboost, speedups on numerical experiments ranges from around 10 to about 1400, at similar predictive-power measured in terms of test-loss.

stat.ME↗