RMTSFMC: A reverse multi-scale time series forecasting method with Mamba and 1D-CNN

Yongdong Zhu

doi:10.1016/j.asoc.2026.116031

RMTSFMC: A reverse multi-scale time series forecasting method with Mamba and 1D-CNN

2026-07-21

0

PRE

AI

DOI：10.1016/j.asoc.2026.116031

Original

Original request for help

Save

Abstract

En 中文

• We introduce a Reverse Encoding mechanism across time scales, achieving diverse temporal resolutions and enriching information at scales of varying resolutions. • We design a Multi-Scale Feature Extractor with 1DCNN and Mamba to extract local and global features, improving temporal modeling and feature extraction. • We design a Future Hybrid Prediction module that projects time-encoded features at different scales onto the same prediction space for complementary outputs.

RMTSFMC: A reverse multi-scale time series forecasting method with Mamba and 1D-CNN

Abstract

Journal

Organization