Search Articles

View query in Help articles search

Search Results (1 to 1 of 1 Results)

Download search results: CSV END BibTex RIS


Leveraging Temporal Trends for Training Contextual Word Embeddings to Address Bias in Biomedical Applications: Development Study

Leveraging Temporal Trends for Training Contextual Word Embeddings to Address Bias in Biomedical Applications: Development Study

We compare our work to the method in the study by Agmon et al [23] in Multimedia Appendix 2. The term “temporal distribution matching” was recently used [25] in an entirely different context: time series forecasting, where given a series of samples and their labels over time, a function from samples to labels is learned. Temporal distribution matching in the context of time series forecasting is a method to handle temporal covariate shifts that harm the performance of the learned prediction model.

Shunit Agmon, Uriel Singer, Kira Radinsky

JMIR AI 2024;3:e49546