Abstract
The invention provides a modular machine learning pipeline that preprocesses structured telemetry logs, applies attention over tokenized key–value pairs, uses a hybrid Neural ODE/SDE-RNN architecture to evolve the embedding over time, employs self-supervised learning for training, supports both offline and online updates, and generates natural language summaries via a language model. Effectively each user’s environment is represented as an embedding which captures the time evolution
Creative Commons License
This work is licensed under a Creative Commons Attribution-Share Alike 4.0 License.
Recommended Citation
INC, HP, "Learning Temporal Embedding of Telemetry via Neural ODEs", Technical Disclosure Commons, (August 26, 2025)
https://www.tdcommons.org/dpubs_series/8511
COinS