DisenTS: Disentangled Channel Evolving Pattern Modeling for Multivariate Time Series Forecasting
Authors:
- Zhiding LiuZLZhiding Liu
- Jiqian YangJYJiqian Yang
- Qingyang MaoQMQingyang Mao
- Yuze ZhaoYZYuze Zhao
- Mingyue ChengMCMingyue Cheng
- Zhi LiZLZhi Li
- Qi LiuQLQi Liu
- Enhong ChenECEnhong Chen
Preprint
Abstract
Multivariate time series forecasting plays a crucial role in various real-world applications. Significant efforts have been made to integrate advanced network architectures and training strategies that enhance the capture of temporal dependencies, thereby improving forecasting accuracy. On the other hand, mainstream approaches typically utilize a single unified model with simplistic channel-mixing embedding or cross-channel attention operations to account for the critical intricate inter-channel dependencies. Moreover, some methods even trade capacity for robust prediction based on the channel-independent assumption. Nonetheless, as time series data may display distinct evolving patterns due to the unique characteristics of each channel (including multiple strong seasonalities and trend changes), the unified modeling methods could yield suboptimal results. To this end, we propose DisenTS, a tailored framework for modeling disentangled channel evolving patterns in general multivariate time series forecasting. The central idea of DisenTS is to model the potential diverse patterns within the multivariate time series data in a decoupled manner. Technically, the framework employs multiple distinct forecasting models, each tasked with uncovering a unique evolving pattern. To guide the learning process without supervision of pattern partition, we introduce a novel Forecaster Aware Gate (FAG) module that generates the routing signals adaptively according to both the forecasters' states and input series' characteristics. The forecasters' states are derived from the Linear Weight Approximation (LWA) strategy, which quantizes the complex deep neural networks into compact matrices. Additionally, the Similarity Constraint (SC) is further proposed to guide each model to specialize in an underlying pattern by minimizing the mutual information between the representations.Keywords
All Related Versions
This publication has 13 references indexed in Scilit:
- The Capacity and Robustness Trade-Off: Revisiting the Channel Independent Strategy for Multivariate Time Series ForecastingIEEE Transactions on Knowledge and Data Engineering, 2024
- Self-Supervised Learning for Time Series Analysis: Taxonomy, Progress, and ProspectsIEEE Transactions on Pattern Analysis and Machine Intelligence, 2024
- DifFormer: Multi-Resolutional Differencing Transformer With Dynamic Ranging for Time Series AnalysisIEEE Transactions on Pattern Analysis and Machine Intelligence, 2023
- Skilful nowcasting of extreme precipitation with NowcastNetNature, 2023
- Learning the Explainable Semantic Relations via Unified Graph Topic-Disentangled Neural NetworksACM Transactions on Knowledge Discovery From Data, 2023
- ST-ExpertNet: A Deep Expert Framework for Traffic PredictionIEEE Transactions on Knowledge and Data Engineering, 2022
- DeepAR: Probabilistic forecasting with autoregressive recurrent networksInternational Journal of Forecasting, 2020
- Forecasting traffic time series with multivariate predicting methodApplied Mathematics and Computation, 2016
- Time series forecasting using a hybrid ARIMA and neural network modelNeurocomputing, 2003
- Estimation procedures for structural time series modelsJournal of Forecasting, 1990