| 1. |
- Allerbo, Oskar, 1985, et al.
(författare)
-
Elastic Gradient Descent, an Iterative Optimization Method Approximating the Solution Paths of the Elastic Net
- 2023
-
Ingår i: Journal of Machine Learning Research. - 1533-7928 .- 1532-4435. ; 24, s. 1-35
-
Tidskriftsartikel (refereegranskat)abstract
- The elastic net combines lasso and ridge regression to fuse the sparsity property of lasso with the grouping property of ridge regression. The connections between ridge regression and gradient descent and between lasso and forward stagewise regression have previously been shown. Similar to how the elastic net generalizes lasso and ridge regression, we introduce elastic gradient descent, a generalization of gradient descent and forward stagewise regression. We theoretically analyze elastic gradient descent and compare it to the elastic net and forward stagewise regression. Parts of the analysis are based on elastic gradient flow, a piecewise analytical construction, obtained for elastic gradient descent with infinitesimal step size. We also compare elastic gradient descent to the elastic net on real and simulated data and show that it provides similar solution paths, but is several orders of magnitude faster. Compared to forward stagewise regression, elastic gradient descent selects a model that, although still sparse, provides considerably lower prediction and estimation errors.
|
|
| 2. |
- Allerbo, Oskar, 1985, et al.
(författare)
-
Flexible, non-parametric modeling using regularized neural networks
- 2022
-
Ingår i: Computational Statistics. - : Springer Science and Business Media LLC. - 0943-4062 .- 1613-9658. ; 37:4, s. 2029-2047
-
Tidskriftsartikel (refereegranskat)abstract
- Non-parametric, additive models are able to capture complex data dependencies in a flexible, yet interpretable way. However, choosing the format of the additive components often requires non-trivial data exploration. Here, as an alternative, we propose PrAda-net, a one-hidden-layer neural network, trained with proximal gradient descent and adaptive lasso. PrAda-net automatically adjusts the size and architecture of the neural network to reflect the complexity and structure of the data. The compact network obtained by PrAda-net can be translated to additive model components, making it suitable for non-parametric statistical modelling with automatic model selection. We demonstrate PrAda-net on simulated data, where we compare the test error performance, variable importance and variable subset identification properties of PrAda-net to other lasso-based regularization approaches for neural networks. We also apply PrAda-net to the massive U.K. black smoke data set, to demonstrate how PrAda-net can be used to model complex and heterogeneous data with spatial and temporal components. In contrast to classical, statistical non-parametric approaches, PrAda-net requires no preliminary modeling to select the functional forms of the additive components, yet still results in an interpretable model representation. © 2021, The Author(s).
|
|
| 3. |
- Allerbo, Oskar, 1985, et al.
(författare)
-
Non-linear, sparse dimensionality reduction via path lasso penalized autoencoders
- 2021
-
Ingår i: Journal of Machine Learning Research. - : Microtome Publishing. - 1532-4435 .- 1533-7928. ; 22
-
Tidskriftsartikel (refereegranskat)abstract
- High-dimensional data sets are often analyzed and explored via the construction of a latent low-dimensional space which enables convenient visualization and efficient predictive modeling or clustering. For complex data structures, linear dimensionality reduction techniques like PCA may not be sufficiently flexible to enable low-dimensional representation. Non-linear dimension reduction techniques, like kernel PCA and autoencoders, suffer from loss of interpretability since each latent variable is dependent of all input dimensions. To address this limitation, we here present path lasso penalized autoencoders. This structured regularization enhances interpretability by penalizing each path through the encoder from an input to a latent variable, thus restricting how many input variables are represented in each latent dimension. Our algorithm uses a group lasso penalty and non-negative matrix factorization to construct a sparse, non-linear latent representation. We compare the path lasso regularized autoencoder to PCA, sparse PCA, autoencoders and sparse autoencoders on real and simulated data sets. We show that the algorithm exhibits much lower reconstruction errors than sparse PCA and parameter-wise lasso regularized autoencoders for low-dimensional representations. Moreover, path lasso representations provide a more accurate reconstruction match, i.e. preserved relative distance between objects in the original and reconstructed spaces. ©2021 Oskar Allerbo and Rebecka Jörnsten.
|
|
| 4. |
- Alm, Jonas, 1984, et al.
(författare)
-
Foreign-currency interest-rate swaps in asset–liability management for insurers
- 2013
-
Ingår i: European Actuarial Journal. - : Springer Science and Business Media LLC. - 2190-9733 .- 2190-9741. ; 3:1, s. 133-158
-
Tidskriftsartikel (refereegranskat)abstract
- We consider an insurer with purely domestic business whose liabilities towards its policy holders have long durations. The relative shortage of domestic government bonds with long maturities makes the insurer’s net asset value sensitive to fluctuations in the zero rates used for liability valuation. Therefore, in order to increase the duration of the insurer’s assets, it is common practice for insurers to take a position as the fixed-rate receiver in an interest-rate swap. We assume that this is not possible in the domestic currency but in a foreign currency supporting a larger market of interest-rate swaps. Monthly data over 16 years are used as the basis for investigating the risks to the future net asset value of the insurer from using foreign-currency interest-rate swaps as a proxy for domestic ones in asset–liability management. We find that although a suitable position in swaps may reduce the standard deviation of the future net asset value it may significantly increase the exposure to tail risk that has a substantial effect on the estimation of the solvency capital requirements.
|
|
| 5. |
- Altay, Özlem, et al.
(författare)
-
Combined Metabolic Activators Accelerates Recovery in Mild-to-Moderate COVID-19
- 2021
-
Ingår i: Advanced Science. - : Wiley. - 2198-3844. ; 8:17
-
Tidskriftsartikel (refereegranskat)abstract
- COVID-19 is associated with mitochondrial dysfunction and metabolic abnormalities, including the deficiencies in nicotinamide adenine dinucleotide (NAD+) and glutathione metabolism. Here it is investigated if administration of a mixture of combined metabolic activators (CMAs) consisting of glutathione and NAD+ precursors can restore metabolic function and thus aid the recovery of COVID-19 patients. CMAs include l-serine, N-acetyl-l-cysteine, nicotinamide riboside, and l-carnitine tartrate, salt form of l-carnitine. Placebo-controlled, open-label phase 2 study and double-blinded phase 3 clinical trials are conducted to investigate the time of symptom-free recovery on ambulatory patients using CMAs. The results of both studies show that the time to complete recovery is significantly shorter in the CMA group (6.6 vs 9.3 d) in phase 2 and (5.7 vs 9.2 d) in phase 3 trials compared to placebo group. A comprehensive analysis of the plasma metabolome and proteome reveals major metabolic changes. Plasma levels of proteins and metabolites associated with inflammation and antioxidant metabolism are significantly improved in patients treated with CMAs as compared to placebo. The results show that treating patients infected with COVID-19 with CMAs lead to a more rapid symptom-free recovery, suggesting a role for such a therapeutic regime in the treatment of infections leading to respiratory problems.
|
|
| 6. |
- Altay, Özlem, et al.
(författare)
-
Current Status of COVID-19 Therapies and Drug Repositioning Applications
- 2020
-
Ingår i: Iscience. - : Elsevier BV. - 2589-0042. ; 23:7
-
Tidskriftsartikel (refereegranskat)abstract
- The rapid and global spread of a new human coronavirus (SARS-CoV-2) has produced an immediate urgency to discover promising targets for the treatment of COVID-19. Drug repositioning is an attractive approach that can facilitate the drug discovery process by repurposing existing pharmaceuticals to treat illnesses other than their primary indications. Here, we review current information concerning the global health issue of COVID-19 including promising approved drugs and ongoing clinical trials for prospective treatment options. In addition, we describe computational approaches to be used in drug repurposing and highlight examples of in silico studies of drug development efforts against SARS-CoV-2.
|
|
| 7. |
- Ames, Ellery, et al.
(författare)
-
Cosmic string and black hole limits of toroidal Vlasov bodies in general relativity
- 2019
-
Ingår i: Physical Review D. - : AMER PHYSICAL SOC. - 2470-0010 .- 2470-0029. ; 99:2
-
Tidskriftsartikel (refereegranskat)abstract
- We numerically investigate limits of a two-parameter family of stationary solutions to the Einstein-Vlasov system. The solutions are toroidal and have nonvanishing angular momentum. As the parameters are tuned to more relativistic solutions (measured e.g., by an increasing redshift) we provide evidence for a sequence of solutions which approaches the extreme Kerr black hole family. Solutions with angular momentum larger than the square of the mass are also investigated, and in the relativistic limit the near-field geometry of such solutions is observed to become locally rotationally symmetric about the matter density. The existence of a deficit angle in these regions is investigated.
|
|
| 8. |
- Andersson, Lars, et al.
(författare)
-
A new tensorial conservation law for Maxwell fields on the Kerr background
- 2017
-
Ingår i: Journal of Differential Geometry. - : International Press of Boston. - 0022-040X .- 1945-743X. ; 105:2, s. 163-176
-
Tidskriftsartikel (refereegranskat)abstract
- A new, conserved, symmetric tensor field for a source-free Maxwell test field on a four-dimensional spacetime with a conformal Killing-Yano tensor, satisfying a certain compatibility condition, is introduced. In particular, this construction works for the Kerr spacetime.
|
|
| 9. |
- Andersson, Lars Erik, et al.
(författare)
-
Decay of solutions to the Maxwell equation on the Schwarzschild background
- 2016
-
Ingår i: Classical and Quantum Gravity. - : IOP Publishing. - 0264-9381 .- 1361-6382. ; 33:8
-
Tidskriftsartikel (refereegranskat)abstract
- A new Morawetz or integrated local energy decay estimate for Maxwell test fields on the exterior of a Schwarzschild black hole spacetime is proved. The proof makes use of a new superenergy tensor H-ab defined in terms of the Maxwell field and its first derivatives. The superenergy tensor, although not conserved, yields a conserved higher order energy current H-ab (partial derivative(t))(b). The tensor Hab vanishes for the static Coulomb field, and the Morawetz estimate proved here therefore yields integrated decay for the Maxwell field to the Coulomb solution on the Schwarzschild exterior.
|
|
| 10. |
- Andersson, Magnus, et al.
(författare)
-
A proposed time-resolved X-ray scattering approach to track local and global conformational changes in membrane transport proteins
- 2008
-
Ingår i: Structure. - : Elsevier BV. - 0969-2126 .- 1878-4186. ; 16:1, s. 21-28
-
Tidskriftsartikel (refereegranskat)abstract
- Time-resolved X-ray scattering has emerged as a powerful technique for studying the rapid structural dynamics of small molecules in solution. Membrane-protein-catalyzed transport processes frequently couple large-scale conformational changes of the transporter with local structural changes perturbing the uptake and release of the transported substrate. Using light-driven halide ion transport catalyzed by halorhodopsin as a model system, we combine molecular dynamics simulations with X-ray scattering calculations to demonstrate how small-molecule time-resolved X-ray scattering can be extended to the study of membrane transport processes. In particular, by introducing strongly scattering atoms to label specific positions within the protein and substrate, the technique of time-resolved wide-angle X-ray scattering can reveal both local and global conformational changes. This approach simultaneously enables the direct visualization of global rearrangements and substrate movement, crucial concepts that underpin the alternating access paradigm for membrane transport proteins.
|
|
