Abstract:Time-series prediction is of high practical value in a wide range of applications such as econometrics and meteorology, where the data are commonly formed by temporal patterns. Most prior works ignore the diversity of dynamic pattern frequency, i.e., different granularities, suffering from insufficient information exploitation. Thus, multi-granularity learning is still under-explored for time-series prediction. In this paper, we propose a Multi-granularity Residual Learning Framework (MRLF) for more effective time series prediction. For a given time series, intuitively, there are more or less semantic overlaps and validity differences among its representations of different granularities. Due to the information redundancy, straightforward methods that leverage multi-granularity data, such as concatenation or ensemble, can easily lead to the model being dominated by the redundant coarse-grained trend information. Therefore, we design a novel residual learning net to model the prior knowledge of the fine-grained data’s distribution through the coarse-grained one. Then, by calculating the residual between multi-granularity data, the redundant information be removed. Furthermore, to alleviate the side effect of validity differences, we introduce a self-supervised objective for confidence estimation, which delivers more effective optimization without the requirement of additional annotation efforts. Extensive experiments on the real-world datasets indicate that multi-granular information significantly improves the time series prediction performance, and our model is superior in capturing such information.

A Multi-output Integration Residual Network for Predicting Time Series Data with Diverse Scales.

A Multiscale Interactive Recurrent Network for Time-Series Forecasting

Multi-Granularity Residual Learning with Confidence Estimation for Time Series Prediction

MSRN-Informer: Time Series Prediction Model Based on Multi-Scale Residual Network

A Deep Multi-Task Representation Learning Method for Time Series Classification and Retrieval.

Multi-scale adaptive attention-based time-variant neural networks for multi-step time series forecasting

TSARNet :Integrated Spatio-Temporal Attention Mechanism for Multivariate Time Series Prediction Deep Learning Framework

HDResNet: Hierarchical-Decomposition Residual Network for Hierarchical Time Series Forecasting.

Multivariate Time Series Forecasting Using Multiscale Recurrent Networks with Scale Attention and Cross-Scale Guidance.

An Intelligent Interconnected Network with Multiple Reservoir Computing

Multi-Stage BiLinear Self-Attention Network for Multi-Sensor Industrial Time Series Classification

DRFormer: Multi-Scale Transformer Utilizing Diverse Receptive Fields for Long Time-Series Forecasting

FormerTime: Hierarchical Multi-Scale Representations for Multivariate Time Series Classification

Time-Aware Multi-Scale RNNs for Time Series Modeling

Multi-resolution Representation with Recurrent Neural Networks Application for Streaming Time Series in IoT.

An Integrated Reservoir Predictor Based on Spatiotemporal Information Transformation

Long Input Sequence Network for Long Time Series Forecasting

MPPN: Multi-Resolution Periodic Pattern Network For Long-Term Time Series Forecasting

Hierarchical Neural Networks for Multivariate Time Series Prediction

WR-IMT: A Time Series Predictive Model for Remaining Useful Life Prediction