ITGPT is a transformer-style architecture for irregular multimodal timeseries that combines SSL and GPT-style training to reach state-of-the-art results on healthcare and predictive maintenance tasks without resampling or imputation.
Recurrent Neural Networks for Multivariate Time Series with Missing Values,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
ITGPT: Generative Pretraining on Irregular Timeseries
ITGPT is a transformer-style architecture for irregular multimodal timeseries that combines SSL and GPT-style training to reach state-of-the-art results on healthcare and predictive maintenance tasks without resampling or imputation.