Tkk Reports in Information and Computer Science Practical Approaches to Principal Component Analysis in the Presence of Missing Values Tkk Reports in Information and Computer Science Practical…

By Alexander Ilin, Tapani Raiko, A. B. T. Korkeakoulu

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Learning Distributed Representations from Reviews for Collaborative Filtering (2015, background): “It has been observed earlier, for instance in [10], that this matrix factorization approach easily overfits the observed ratings, leading to poor generalization performance on the held-out set, or unseen user-product pairs.”
missMDA: A Package for Handling Missing Values in Multivariate Data Analysis (2016, background): “However, Josse et al. (2009) and Ilin and Raiko (2010) highlighted this algorithm’s overfitting problems, and suggested regularized versions to tackle these issues.”
Self-organization and missing values in SOM and GTM (2015, methodology): “…99 (10%, SOM init), 9 (10%, PCA init), 4 (30%), 3 (50%, PCA init), 4 (50%, SOM init) map units for the GTM. Topographic mappings are compared with naive mean imputation and Variational Bayesian PCA (VBPCA) [35], which can be used as a generic black box imputation even with extremely sparse data.”
Active Sampler: Light-weight Accelerator for Complex Data Analytics at Scale (2015, unknown): “…many advanced large-scale machine learning models have been devised, from million-dimension linear models (e.g. Logistic Regression [9], Support Vector Machine [25], feature selection [16, 31], Principal Component Analysis [10]) to complex models like Deep Neural Networks [4] or topic models [8].”
Global analytic solution of fully-observed variational Bayesian matrix factorization (2013, methodology): “In practice, the VBMF solution is computed by the iterated conditional modes (ICM) algorithm (Besag, 1986; Bishop, 2006), where the mean and the covariance of the posterior distributions are iteratively updated until convergence (Lim and Teh, 2007; Ilin and Raiko, 2010).”
Bayesian Canonical correlation analysis (2013, methodology): “We also assume fully observed data; techniques similar to what Ilin and Raiko (2010) propose for Bayesian PCA could be adopted to handle missing data.”
Ecomorphological diversifications of Mesozoic marine reptiles: the roles of ecological opportunity and extinction (2016, methodology): “Missing values were accounted for using iterative imputation with 10,000 bootstrap replicates (Ilin and Raiko 2010).”