Time Series Classification (TSC) is an important and challenging problem in data mining. With the increase of time series data availability, hundreds of TSC algorithms have been proposed. Among these methods, only a few have considered Deep Neural Networks (DNNs) to perform this task. This is surprising as deep learning has seen very successful applications in the last years. DNNs have indeed revolutionized the field of computer vision especially with the advent of novel deeper architectures such as Residual and Convolutional Neural Networks. Apart from images, sequential data such as text and audio can also be processed with DNNs to reach state-of-the-art performance for document classification and speech recognition. In this article, we study the current state-of-the-art performance of deep learning algorithms for TSC by presenting an empirical study of the most recent DNN architectures for TSC. We give an overview of the most successful deep learning applications in various time series domains under a unified taxonomy of DNNs for TSC. We also provide an open source deep learning framework to the TSC community where we implemented each of the compared approaches and evaluated them on a univariate TSC benchmark (the UCR/UEA archive) and 12 multivariate time series datasets. By training 8730 deep learning models on 97 time series datasets, we propose the most exhaustive study of DNNs for TSC to date.

https://link.springer.com/content/pdf/10.1007/s10618-019-00619-1.pdf

Deep learning for time series classification: a review

A deep convolutional neural network model to classify heartbeats

Regardless of successful applications of the convolutional neural networks (CNNs) in different fields, its application to seismic waveform classification and first-break (FB) picking has not been explored yet. This letter investigates the application of CNNs for classifying time-space waveforms from seismic shot gathers and picking FBs of both direct wave and refracted wave. We use representative subimage samples with two types of labeled waveform classification to supervise CNNs training. The goal is to obtain the optimal weights and biases in CNNs, which are solved by minimizing the error between predicted and target label classification. The trained CNNs can be utilized to automatically extract a set of time-space attributes or features from any subimage in shot gathers. These attributes are subsequently inputted to the trained fully connected layer of CNNs to output two values between 0 and 1. Based on the two-element outputs, a discriminant score function is defined to provide a single indication for classifying input waveforms. The FB is then located from the calculated score maps by sequentially using a threshold, the first local minimum rule of every trace and a median filter. Finally, we adopt synthetic and real shot data examples to demonstrate the effectiveness of CNNs-based waveform classification and FB picking. The results illustrate that CNN is an efficient automatic data-driven classifier and picker.

/pdf/seismic-waveform-classification-and-first-break-picking-1776zxtuxs.pdf

Seismic Waveform Classification and First-Break Picking Using Convolution Neural Networks

https://nottingham-repository.worktribe.com/preview/2315217/Multi_Time_Series_Anomaly_Detection_for_Industry_4_0__Neurocomputing_.pdf

Multi-head CNN–RNN for multi-time series anomaly detection: An industrial case study

Statistical and machine learning models in credit scoring: A systematic literature survey

Convolutional neural networks for time series classification

The micromotion dynamics and geometrical shape are considered to be essential characteristics for exoatmospheric targets discrimination. Many methods have been investigated to retrieve the micromotion features using radar signals returned from targets of a given shape. We explore a way of jointly estimating micromotion dynamics and geometrical shape parameters from the infrared (IR) signals of targets in remote detection distance. It is found that the micromotion dynamics of the target would induce a periodic fluctuating variation on the IR irradiance intensity signature. In addition to the micromotion characteristics, the fluctuation could also reflect target structure properties, which offer a possible clue in extracting the features of micromotion dynamics and geometrical shape. Thus, the data model of target IR irradiance intensity signatures induced by micromotion patterns including spinning, coning, and tumbling is developed, and a method of parameters estimation based on joint optimization analysis techniques is proposed. Experimental results demonstrated that the parameters of target micromotion dynamics and geometrical shape can be effectively estimated using the proposed method, if the input signature contains multiple dominant frequency components.

Micromotion dynamics and geometrical shape parameters estimation of exoatmospheric infrared targets

A novel waveforms classification method based on convolutional neural networks (CNN) is proposed in this paper. Firstly, convolution and pooling operations are cross used for generating deep features, and then fully connected to the output layer for classification. Different from other traditional approaches which need human-designed features, CNN can discover and extract the suitable internal structure of the input waveform to obtain deep features for classification automatically. So that the generalization ability of this method is significantly improved comparing to other methods. Experimental results show that CNN can obtain state of the art performance for waveforms classification in terms of classification accuracy and noise tolerance.

Waveforms classification based on convolutional neural networks

The wave data measured by CFOSAT (China France Oceanography Satellite) have been validated mainly based on numerical model outputs and altimetry products on a global scale. It is still necessary to further calibrate the data for specific regions, e.g., the southern South China Sea. This study analyses the practicability of calibrating the dominant wavelength by using artificial neural networks and mean impact value analysis based on two sets of buoy data with a 2-year observation period and contemporaneous ERA5 reanalysis data. The artificial neural network modeling experiments are repeated 1000 times randomly by Monte Carlo methods to avoid sampling uncertainty. Both experimental results based on the random sampling method and chronological sampling method are performed. Independent buoy observations are used to validate the calibration model. The results show that although there are obvious differences between the CFOSAT wavelength data and the field observations, the parameters observed by the satellite itself can effectively calibrate the data. In addition to the wavelength, nadir significant wave height, nadir wind speed, and the distance between the calibration point and satellite observation point are the most important parameters for the calibration. Accurate data from other sources, such as ERA5, would be helpful to further improve the calibration results. The variable contributing the most to the calibration effect is the mean wave period, which virtually provides relatively accurate wavelength information for the calibration network. These results verify the possibility of synchronous self-calibration for the CFOSAT wavelength data and provide a reference for the further calibration of the satellite products in other regions. 

/pdf/calibration-experiments-of-cfosat-wavelength-in-the-southern-1yqebi4q.pdf

Calibration Experiments of CFOSAT Wavelength in the Southern South China Sea by Artificial Neural Networks

Micro-motion dynamics and geometrical shape are considered to be essential evidence for infrared (IR) ballistic target recognition. However, it is usually hard or even impossible to describe the geometrical shape of an unknown target with a finite number of parameters, which results in a very difficult task to estimate target micro-motion parameters from the IR signals. Considering the shapes of ballistic targets are relatively simple, this paper explores a joint optimization technique to estimate micro-motion and dominant geometrical shape parameters from sparse decomposition representation of IR irradiance intensity signatures. By dividing an observed target surface into a number of segmented patches, an IR signature of the target can be approximately modeled as a linear combination of the observation IR signatures from the dominant segmented patches. Given this, a sparse decomposition representation of the IR signature is established with the dictionary elements defined as each segmented patch’s IR signature. Then, an iterative optimization method, based on the batch second-order gradient descent algorithm, is proposed to jointly estimate target micro-motion and geometrical shape parameters. Experimental results demonstrate that the micro-motion and geometrical shape parameters can be effectively estimated using the proposed method, when the noise of the IR signature is in an acceptable level, for example, SNR>0  dB.

Junliang Liu

Papers

Convolutional neural networks for time series classification

Micromotion dynamics and geometrical shape parameters estimation of exoatmospheric infrared targets

Waveforms classification based on convolutional neural networks

Calibration Experiments of CFOSAT Wavelength in the Southern South China Sea by Artificial Neural Networks

Ballistic targets micro-motion and geometrical shape parameters estimation from sparse decomposition representation of infrared signatures