osiris
Interpretable Generative and Discriminative Learning for Multimodal and Incomplete Clinical Data
Belenguer-Llorens, Albert, Sevilla-Salcedo, Carlos, Mourao-Miranda, Janaina, Gómez-Verdejo, Vanessa
Real-world clinical problems are often characterized by multimodal data, usually associated with incomplete views and limited sample sizes in their cohorts, posing significant limitations for machine learning algorithms. In this work, we propose a Bayesian approach designed to efficiently handle these challenges while providing interpretable solutions. Our approach integrates (1) a generative formulation to capture cross-view relationships with a semi-supervised strategy, and (2) a discriminative task-oriented formulation to identify relevant information for specific downstream objectives. This dual generative-discriminative formulation offers both general understanding and task-specific insights; thus, it provides an automatic imputation of the missing views while enabling robust inference across different data sources. The potential of this approach becomes evident when applied to the multimodal clinical data, where our algorithm is able to capture and disentangle the complex interactions among biological, psychological, and sociodemographic modalities.
State Relevance for Off-Policy Evaluation
Shen, Simon P., Ma, Yecheng Jason, Gottesman, Omer, Doshi-Velez, Finale
Importance sampling-based estimators for off-policy evaluation (OPE) are valued for their simplicity, unbiasedness, and reliance on relatively few assumptions. However, the variance of these estimators is often high, especially when trajectories are of different lengths. In this work, we introduce Omitting-States-Irrelevant-to-Return Importance Sampling (OSIRIS), an estimator which reduces variance by strategically omitting likelihood ratios associated with certain states. We formalize the conditions under which OSIRIS is unbiased and has lower variance than ordinary importance sampling, and we demonstrate these properties empirically.
Xbox Live Games With Gold May 2017: Free Games Include 'Tomb Raider,' 'Star Wars' Titles
For May, Xbox Live Gold members will get a mix of free indie, major and "Star Wars" titles. Microsoft's Major Nelson says May's lineup of free games from Xbox Live's Games With Gold program is led on the Xbox One by "Giana Sisters: Twisted Dreams -- Director's Cut." The platformer will be available for the whole month of May and will get paired with "Lara Croft and the Temple of Osiris" May 16-June 15. On the Xbox 360, the Games With Gold calendar starts with "Star Wars: The Force Unleashed II," which will be available from May 1 to May 15. The game will be followed by "Lego Star Wars: The Complete Saga," which runs from May 16 to May 31.