Johan Larsson, postdoc
Johan Larsson is employed as a postdoc on a Thiele Data Science Fellowship from 3 September 2024, working in the Statistics and Probability section.
Johan has a doctoral degree from the Department of Statistics, Lund University, Sweden, and presented his thesis “Optimization and Algorithms in Sparse Regression” in July this year, supervised by Jonas Wallin and Malgorzata Bogdan.
His research has focused on regularized methods such as the lasso and Sorted L-One Penalized Estimation (SLOPE), in particular, numerical algorithms and optimization methods for these methods to improve the speed at which they are fit. His results have included so-called screening rules for the lasso and SLOPE, as well as a novel coordinate descent algorithm for SLOPE. Recently he has been studying the effects of feature (predictor) normalization in regularized regression.
In Copenhagen, Johan will be working with Niels Richard Hansen.
You can meet Johan at the office 04.3.06 and his homepage is found at https://jolars.co.