Self-Supervised Transformer for Sparse and Irregularly Sampled Multivariate Clinical Time-Series
Self-Supervised Transformer for Sparse and Irregularly Sampled Multivariate Clinical Time-Series
Multivariate time-series data are frequently observed in critical care settings and are typically characterized by sparsity (missing information) and irregular time intervals. Existing approaches for learning representations in this domain handle these challenges by either aggregation or imputation of values, which in-turn suppresses the fine-grained information and adds undesirable noise/overhead …