change loss function - compare to benchmark datasets

Question

change loss function - compare to benchmark datasets

cheind opened this issue 3 years ago · 4 comments

cheind commented 3 years ago

https://proceedings.neurips.cc/paper/2019/file/466accbac9a66b805ba50e42ad715740-Paper.pdf

Answer 1 · 2021-11-04T08:47:36.000Z

https://arxiv.org/pdf/2104.04610.pdf

Answer 2 · 2021-11-04T18:11:29.000Z

ecg setup https://arxiv.org/pdf/1703.01541.pdf

Multistep-ahead prediction experiments
In this section, we present preliminary experiments for the
task of multistep-ahead prediction, described in §3.4.
Experimental setup. We use the training and test sets predefined in the UCR archive. In both the training and test
sets, we use the first 60% of the time series as input and the
remaining 40% as output, ignoring class information

Answer 3 · 2021-11-04T18:50:26.000Z

electricity: use hourly dataset https://zenodo.org/record/3889830#.YYQqh2DMIuU

Answer 4 · 2021-12-09T18:46:23.000Z

using mnist as a 2d example instead