|
- ModernTCN: A Modern Pure Convolution Structure for General Time Series . . .
The paper presents a pure convolution architecture for time series analysis The proposed ModernTCN block is partially inspired by modern convolution blocks as in ConvNeXt ModernTCN mainly utilizes Depthwise Convolution and Grouped Pointwise convolution to learn cross-variable and cross-feature information in a decoupled way
- M TCN: A M P CONVOLUTION STRUCTURE FOR S ANALYSIS - OpenReview
As the outcome, we propose a modern pure convolution structure, namely ModernTCN, to efficiently utilize cross-time and cross-variable dependency for general time series analysis We evaluate ModernTCN on five mainstream analysis tasks, including long-term and short-term forecasting, imputation, classification and anomaly detection
- ModernTCN Revisited: A Critical Look at the Experimental Setup in . . .
While numerous time series models claim state-of-the-art performance, their evaluation often relies on flawed experimental setups, leading to questionable conclusions This study provides a
- ModernTCNRevisited: ACriticalLookattheExperimental . . .
ModernTCN is a general time series analysis model designed to enhance performance across five key time series tasks: long-term forecasting, short-term forecasting, classification, imputation, and anomaly detection It modernizes traditional Temporal Convolutional Networks (TCNs) Bai et al (2018) by enlarging the effective receptive field (ERF), drawing inspiration from computer vision
- Revisions | OpenReview
As a pure convolution structure, ModernTCN still achieves the consistent state-of-the-art performance on five mainstream time series analysis tasks while maintaining the efficiency advantage of convolution-based models, therefore providing a better balance of efficiency and performance than state-of-the-art Transformer-based and MLP-based models
- Xue Wang - OpenReview
Promoting openness in scientific communication and the peer-review process
- DeformableTST: Transformer for Time Series Forecasting without. . .
With the proposal of patching technique in time series forecasting, Transformerbased models have achieved compelling performance and gained great interest from the time series community But at the
- Test Time Learning for Time Series Forecasting
ModernTCN employs depthwise-separable convolutions to process time series data eficiently Let Cin and Cout denote the input and output channel dimensions, and k the kernel size
|
|
|