Long-term series forecasting with Query Selector – efficient model of sparse attention

07/19/2021

∙

by   Jacek Klimek, et al.

∙

1

∙

share

Various modifications of TRANSFORMER were recently used to solve time-series forecasting problem. We propose Query Selector - an efficient, deterministic algorithm for sparse attention matrix. Experiments show it achieves state-of-the art results on ETT data set.