Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

ContiFormer: Continuous-Time Transformer for Irregular Time Series Modeling

About

Modeling continuous-time dynamics on irregular time series is critical to account for data evolution and correlations that occur continuously. Traditional methods including recurrent neural networks or Transformer models leverage inductive bias via powerful neural architectures to capture complex patterns. However, due to their discrete characteristic, they have limitations in generalizing to continuous-time data paradigms. Though neural ordinary differential equations (Neural ODEs) and their variants have shown promising results in dealing with irregular time series, they often fail to capture the intricate correlations within these sequences. It is challenging yet demanding to concurrently model the relationship between input data points and capture the dynamic changes of the continuous-time system. To tackle this problem, we propose ContiFormer that extends the relation modeling of vanilla Transformer to the continuous-time domain, which explicitly incorporates the modeling abilities of continuous dynamics of Neural ODEs with the attention mechanism of Transformers. We mathematically characterize the expressive power of ContiFormer and illustrate that, by curated designs of function hypothesis, many Transformer variants specialized in irregular time series modeling can be covered as a special case of ContiFormer. A wide range of experiments on both synthetic and real-world datasets have illustrated the superior modeling capacities and prediction performance of ContiFormer on irregular time series data. The project link is https://seqml.github.io/contiformer/.

Yuqi Chen, Kan Ren, Yansen Wang, Yuchen Fang, Weiwei Sun, Dongsheng Li• 2024

Related benchmarks

TaskDatasetResultRank
Time Series ForecastingETTh2 (test)
MSE0.356
140
Time Series ForecastingWeather (test)
MSE0.257
110
Event PredictionStackOverflow
RMSE0.948
42
Univariate Time Series ForecastingExchange (test)
MSE0.339
38
Degradation EstimationPRONOSTIA
MSE37.82
33
Lane-Keeping Trajectory PredictionUdacity Simulator
MSE0.0174
33
Irregular Time Series ClassificationE-MNIST
Accuracy96.04
33
Lane-Keeping Action ClassificationOpenAI CarRacing
Accuracy80.47
33
Degradation EstimationXJTU-SY
MSE34.71
33
Degradation EstimationHUST
MSE43.81
33
Showing 10 of 37 rows

Other info

Code

Follow for update