A Time Series is Worth 64 Words: Long-term Forecasting with Transformers
Explore this paper's citation graph
Summary
The channel-independent patch time series Transformer (PatchTST) can improve the long-term forecasting accuracy significantly when compared with that of SOTA Transformer-based models and applies to self-supervised pre-training tasks and attain excellent fine-tuning performance.
- Type
- preprint
- Published
- 2022-11-27
- Cited by
- 4,406
- References
- 46
- Access
- Open access
- OpenAlex
- https://openalex.org/W4310416691
- Semantic Scholar
- https://api.semanticscholar.org/CorpusID:254044221
Keywords
Computer science, Transformer, Embedding, Univariate, Segmentation
References
- Time series analysis, forecasting and control
- Exchange Rate Predictability
- Long Short-Term Memory
- Japanese and Korean voice search
- Instance Normalization: The Missing Ingredient for Fast Stylization
- DeepAR: Probabilistic Forecasting with Autoregressive Recurrent Networks
- An Empirical Evaluation of Generic Convolutional and Recurrent Networks for Sequence Modeling
- Gaussian Error Linear Units (GELUs)
- Unsupervised Scalable Representation Learning for Multivariate Time Series
- BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
- Enhancing the Locality and Breaking the Memory Bottleneck of Transformer on Time Series Forecasting
- A Comparative Study on Transformer vs RNN in Speech Applications
- Time-series forecasting with deep learning: a survey
- A Transformer-based Framework for Multivariate Time Series Representation Learning
- An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
- wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations
- Spectral Temporal Graph Neural Network for Multivariate Time-series Forecasting
- An Experimental Review on Deep Learning Architectures for Time Series Forecasting
- Deep Learning for Time Series Forecasting: A Survey
- Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting
Cited by
- DynaConF: Dynamic Forecasting of Non-Stationary Time-Series
- TILDE-Q: A Transformation Invariant Loss Function for Time-Series Forecasting
- SimMTM: A Simple Pre-Training Framework for Masked Time-Series Modeling
- An Accurate and Interpretable Framework for Trustworthy Process Monitoring
- Power Time Series Forecasting by Pretrained LM
- Your time series is worth a binary image: machine vision assisted deep framework for time series forecasting
- TSMixer: An all-MLP Architecture for Time Series Forecasting
- Towards Better Dynamic Graph Learning: New Architecture and Unified Library
- Time Series as Images: Vision Transformer for Irregularly Sampled Time Series
- Long-term Forecasting with TiDE: Time-series Dense Encoder
- Time Series Analysis Based on Informer Algorithms: A Survey
- Mlinear: Rethink the Linear Model for Time-series Forecasting
- Revisiting long-term time series forecasting: an investigation on affine mapping
- Disentangling Structured Components: Towards Adaptive, Interpretable and Scalable Time Series Forecasting
- CARD: Channel Aligned Robust Blend Transformer for Time Series Forecasting
- A Joint Time-frequency Domain Transformer for Multivariate Time Series Forecasting
- FDNet: Focal Decomposed Network for Efficient, Robust and Practical Time Series Forecasting
- Long-term Wind Power Forecasting with Hierarchical Spatial-Temporal Transformer
- Koopa: Learning Non-stationary Time Series Dynamics with Koopman Predictors
- Improving *day-ahead* Solar Irradiance Time Series Forecasting by Leveraging Spatio-Temporal Context
Related papers
- Comparing multivariate and univariate subject-specific reference regions for blood constituents in healthy persons.
- Theory and practice of multivariate arma forecasting
- Outliers in multivariate time series
- A multivariate approach to a specific problem of grouping maize cultivars.
- Multivariate trend testing of lake water quality
- Forecasting precious metal returns with multivariate random forests
- When univariate model-free time series prediction is better than multivariate
- Multivariate rank-based forecast combining techniques
- Forecasting of Human Development Index of Latin American Countries Through Data Mining Techniques
- New multivariate orderings based on conditional distributions