Multi-Scale Convolutional Neural Networks for Time Series Classification

Abstract

Time series classification (TSC), the problem of predicting class labels of time series, has been around for decades within the community of data mining and machine learning, and found many important applications such as biomedical engineering and clinical prediction. However, it still remains challenging and falls short of classification accuracy and efficiency. Traditional approaches typically involve extracting discriminative features from the original time series using dynamic time warping (DTW) or shapelet transformation, based on which an off-the-shelf classifier can be applied. These methods are ad-hoc and separate the feature extraction part with the classification part, which limits their accuracy performance. Plus, most existing methods fail to take into account the fact that time series often have features at different time scales. To address these problems, we propose a novel end-to-end neural network model, Multi-Scale Convolutional Neural Networks (MCNN), which incorporates feature extraction and classification in a single framework. Leveraging a novel multi-branch layer and learnable convolutional layers, MCNN automatically extracts features at different scales and frequencies, leading to superior feature representation. MCNN is also computationally efficient, as it naturally leverages GPU computing. We conduct comprehensive empirical evaluation with various existing methods on a large number of benchmark datasets, and show that MCNN advances the state-of-the-art by achieving superior accuracy performance than other leading methods.

Cited by in corpus (59)

• Deep learning for time series classification: a review
• LSTM Fully Convolutional Networks for Time Series Classification
• Applications of deep learning in stock market prediction: recent progress
• Self-attention for raw optical Satellite Time Series Classification
• Multi-task Self-Supervised Learning for Human Activity Detection
• Transfer learning for time series classification
• SCINet: Time Series Modeling and Forecasting with Sample Convolution and Interaction
• Unsupervised Scalable Representation Learning for Multivariate Time Series
• Insights into LSTM Fully Convolutional Networks for Time Series Classification
• Time Series Anomaly Detection Using Convolutional Neural Networks and Transfer Learning
• Gated Transformer Networks for Multivariate Time Series Classification
• Deep Neural Network Ensembles for Time Series Classification
• Deep Gated Recurrent and Convolutional Network Hybrid Model for Univariate Time Series Classification
• Classification of Time-Series Images Using Deep Convolutional Neural Networks
• AR-Net: A simple Auto-Regressive Neural Network for time-series
• Time Series Classification from Scratch with Deep Neural Networks: A Strong Baseline
• Transformer Encoder with Multiscale Deep Learning for Pain Classification Using Physiological Signals
• Transfer Learning and SpecAugment applied to SSVEP Based BCI Classification
• Classification with 2-D Convolutional Neural Networks for breast cancer diagnosis
• Discrete Event, Continuous Time RNNs
• Earliness-Aware Deep Convolutional Networks for Early Time Series Classification
• dCAM: Dimension-wise Class Activation Map for Explaining Multivariate Data Series Classification
• Cost-Sensitive Convolution based Neural Networks for Imbalanced Time-Series Classification
• Learning to Exploit Invariances in Clinical Time-Series Data using Sequence Transformer Networks
• Deep reinforcement learning for time series: playing idealized trading games
• Classification of Multiwavelength Transients with Machine Learning
• TDEFSI: Theory Guided Deep Learning Based Epidemic Forecasting with Synthetic Information
• Multi-Scale RCNN Model for Financial Time-series Classification
• Autoregressive Convolutional Recurrent Neural Network for Univariate and Multivariate Time Series Prediction
• Autoencoder-based time series clustering with energy applications
• Deep learning for time series classification
• Towards data-driven stroke rehabilitation via wearable sensors and deep learning
• Surrogate Optimization of Deep Neural Networks for Groundwater Predictions
• DTW-Merge: A Novel Data Augmentation Technique for Time Series Classification
• Multilevel Wavelet Decomposition Network for Interpretable Time Series Analysis
• Feature Selection for Multivariate Time Series via Network Pruning
• Deep neural network for solving differential equations motivated by Legendre-Galerkin approximation
• Deep Ensemble Tensor Factorization for Longitudinal Patient Trajectories Classification
• SincVAE: A new semi-supervised approach to improve anomaly detection on EEG data using SincNet and variational autoencoder
• Correlative Channel-Aware Fusion for Multi-View Time Series Classification
• Merchant Category Identification Using Credit Card Transactions
• A Hybrid Deep Learning Model for Predictive Flood Warning and Situation Awareness using Channel Network Sensors Data
• Wearable data from subjects playing Super Mario, sitting university exams, or performing physical exercise help detect acute mood episodes via self-supervised learning
• EnK: Encoding time-information in convolution
• Time Scale Network: A Shallow Neural Network For Time Series Data
• ShortFuse: Biomedical Time Series Representations in the Presence of Structured Information
• Learning Correlation Space for Time Series
• Ballroom Dance Movement Recognition Using a Smart Watch
• SocialGrid: A TCN-enhanced Method for Online Discussion Forecasting
• Modeling Conceptual Characteristics of Virtual Machines for CPU Utilization Prediction
• A Daily Tourism Demand Prediction Framework Based on Multi-head Attention CNN: The Case of The Foreign Entrant in South Korea
• Bifurcation Spiking Neural Network
• Accuracy Improvement for Fully Convolutional Networks via Selective Augmentation with Applications to Electrocardiogram Data
• GeoStat Representations of Time Series for Fast Classification
• Multi-Faceted Representation Learning with Hybrid Architecture for Time Series Classification
• Spectrum Attention Mechanism for Time Series Classification
• Optimizing Hyperparameters in CNNs using Bilevel Programming in Time Series Data
• Dynamic Time Warp Convolutional Networks
• Piece-wise Matching Layer in Representation Learning for ECG Classification