AI Paper Atlas

Research map / Time series / Foundation models for time series

SimMTM: A Simple Pre-Training Framework for Masked Time-Series Modeling

Jiaxiang Dong, Haixu Wu, Haoran Zhang, Li Zhang, Jianmin Wang, Mingsheng Long

NeurIPS 2023 · Time series › Foundation models for time series · approach: Time series representation learning · 40 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
SimMTM pre-training vs random initialization (ablation)Transformer forecasting · Forecasting in-domain avgMSE: 0.343 -> 0.308Improved
SimMTM vs Ti-MAE pre-trainingforecasting · Forecasting in-domain avgMSE: 0.336 -> 0.308Improved
SimMTM vs TF-C pre-trainingclassification · SleepEEG -> EMGaccuracy (%): 81.74 -> 97.56Improved
TST pre-training vs random init (masked baseline)classification · Epilepsy -> Epilepsyaccuracy (%): 89.83 -> 80.89Hurt
add reconstruction loss and constraint loss (full SimMTM) vs random init (ablation)SimMTM · SleepEEG -> EMGaccuracy (%): +9.7% and +16.0% absoluteImproved
add constraint loss (manifold structure in series-wise space) (ablation)SimMTM · SleepEEG -> EMGaccuracy (%): +16.0% absoluteImproved
aggregate negatives too (PNSA) instead of positives only (PSA) (ablation)SimMTM · ETTh1 in-domainMSE: 0.420 -> 0.409Improved
aggregate negatives too (PNSA) instead of positives only (PSA) (ablation)SimMTM · SleepEEG -> EMGaccuracy (%): 87.80 -> 97.56Improved
add SimMTM pre-training to PatchTSTPatchTST · ETTh1MSE: 0.417 -> 0.409Improved
add PatchTST sub-series masking pre-trainingPatchTST · ETTh1MSE: 0.417 -> 0.430Hurt
use multiple masked series (M>1) instead of one (M=1) (ablation)SimMTM · ETTh1 input-336-predict-96MSEImproved
geometric masking instead of random masking (ablation)SimMTM · ETTh1MSEImproved

Later papers that compared with this method

6 later accepted papers report SimMTM as a baseline in their results tables; 5 of them report a better number on most of the results they share with it.

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy