Multi-Scale Convolutional Neural Networks for Time Series Classification

By Zhicheng Cui, Yixin Chen. 2016

622 citations recorded by the source corpus.

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.

Subfields: Artificial Intelligence, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Signal Processing

Citations in the corpus

An End-to-End Deep Learning Architecture for Graph Classification (2018, background): “…few years have seen the growing prevalence of neural networks on application domains such as image classification (Alex, Sutskever, and Hinton 2012), natural language processing (Mikolov et al. 2013), reinforcement learning (Mnih et al. 2013), and time series analysis (Cui, Chen, and Chen 2016).”
InceptionTime: Finding AlexNet for time series classification (2019, neutral): “…of deep learning techniques in supervised learning such as image recognition (Zhang et al., 2018) and natural language processing (Guan et al., 2019), researchers started investigating these complex machine learning models for TSC (Wang et al., 2017; Cui et al., 2016; Ismail Fawaz et al., 2019a).”
ROCKET: exceptionally fast and accurate time series classification using random convolutional kernels (2019, methodology): “The convolution operation is based on the dot product which, if used in conjunctionwith global max pooling, is very similar to Euclidean distance typically used in shapelet methods (Cui et al. 2016).”
Data Augmentation for Time Series Classification using Convolutional Neural Networks (2016, methodology): “In [6], authors propose different data-”
Predicting mortgage default using convolutional neural networks (2018, methodology): “Cui et al. (2016) propose a sliding window technique where they split the time series in many overlapping slices, and train a classifier using each slice as an individual series.”
Applications of deep learning in stock market prediction: recent progress (2020, methodology): “, 2017), SFM RNN (Hu & Qi, 2017), MS-CNN (Cui et al., 2016) Jin et al.”
Multi-task Self-Supervised Learning for Human Activity Detection (2019, methodology): “[12] used transformed signals as extra channels to the model for learning multiscale features.”
Multiscale Convolutional Neural Networks for Fault Diagnosis of Wind Turbine Gearbox (2019, methodology): “In a recent study in [30], a similar MCNN was proposed to address the time series classification issue and has shown the state-of-the-art performance on a large number of benchmark data sets.”
Multi-input CNN-GRU based human activity recognition using wearable sensors (2021, methodology): “MCNN [19], yet another deep learning approach that designed an end-to-end classifier for univariate time series data, made use of CNN.”
An Automated System for Epilepsy Detection using EEG Brain Signals based on Deep Learning Approach (2018, methodology): “Recently, 1D-CNN has been successfully used for text understanding, music generation, and other time series data (Cui et al., 2016; Ince et al., 2016; LeCun et al., 1998; X. Zhang et al., 2015).”
Deep Learning Based Forecasting of Critical Infrastructure Data (2017, methodology): “[8].”
Deep Learning for ECG Analysis: Benchmarks and Insights from PTB-XL (2020, methodology): “During training, we follow the sliding window approach that is commonly used in time series classiﬁcation, see e.g. [24], [26], [49], [59].”
Time Series Data Augmentation for Deep Learning: A Survey (2020, background): “Introduced in [ Cui et al. , 2016 ] , window cropping is similar to cropping in CV area.”