Rough Transformers: Lightweight and Continuous Time Series Modelling through Signature Patching Fernando Moreno-Pino 1, Álvaro Arroyo 1,2, Harrison Waldon 1, Xiaowen Dong

Feb-17-2026, 22:02:42 GMT–Neural Information Processing Systems 

In many real-world scenarios, sequential data are time-series sampled from some underlying continuous-time process, so datasets consist of long, irregularly sampled sequences of varied lengths.