Blar i NTNU Open på forfatter "Johnsen, Pål Vegard"
-
Explainability and validity of statistical methods for genome-wide association studies: Extending Shapley-based explanation methods and adapting saddlepoint approximations
Johnsen, Pål Vegard (Doctoral theses at NTNU;2022:26, Doctoral thesis, 2022) -
Inferring feature importance with uncertainties with application to large genotype data
Johnsen, Pål Vegard; Strumke, Inga; Langaas, Mette; DeWan, Andrew Thomas; Riemer-Sørensen, Signe (Peer reviewed; Journal article, 2023)Estimating feature importance, which is the contribution of a prediction or several predictions due to a feature, is an essential aspect of explaining data-based models. Besides explaining the model itself, an equally ... -
Investigating the Capability of Generative Adversarial Networks of Capturing Implicit Laws in Physical Systems
Kohmann, Eivind (Master thesis, 2023)Utfordringen med tolkning av generative modeller, spesielt blant Generative Adversarial Networks (GANs), utgjør et betydelig problem når det handler om å forstå hva de faktisk lærer. Til tross for deres bemerkelsesverdige ... -
Saddlepoint approximations to score test statistics in logistic regression for analyzing genome-wide association studies
Johnsen, Pål Vegard; Bakke, Øyvind; Bjørnland, Thea; DeWan, Andrew Thomas; Langaas, Mette (Peer reviewed; Journal article, 2023)We investigate saddlepoint approximations of tail probabilities of the score test statistic in logistic regression for genome-wide association studies. The inaccuracy in the normal approximation of the score test statistic ... -
Stochastic modelling and analysis of neutral tumor evolution. - Can neutral tumor evolution be inferred from real data?
Johnsen, Pål Vegard (Master thesis, 2018)We develop a branching process with accumulating mutations to model neutral tumor evolution. Furthermore, by using this model, it will be shown in details which approximations and simplifications that are necessary in order ...