Data Augmentation techniques in time series domain: a survey and taxonomy
Open accessThe DOI matches its Crossref record; 103 of the 104 references it cites were found in a registry or an index.
Abstract
With the latest advances in deep learning-based generative models, it has not taken long to take advantage of their remarkable performance in the area of time series. Deep neural networks used to work with time series heavily depend on the size and consistency of the datasets used in training. These features are not usually abundant in the real world, where they are usually limited and often have constraints that must be guaranteed. Therefore, an effective way to increase the amount of data is by using data augmentation techniques, either by adding noise or permutations and by generating new synthetic data. This work systematically reviews the current state of the art in the area to provide an overview of all available algorithms and proposes a taxonomy of the most relevant research. The efficiency of the different variants will be evaluated as a central part of the process, as well as the different metrics to evaluate the performance and the main problems concerning each model will be analysed. The ultimate aim of this study is to provide a summary of the evolution and performance of areas that produce better results to guide future researchers in this field.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Figures





![Fig. 6 GlowImp architecture. Based on the figure of [89]](https://figures.cochat.ai/paper-figures/98bcd42a33a84ccba78c91ea5f2ffe31/data-augmentation-techniques-in-time-series-domain-a-survey-and-taxonomy-figure-6.png)


![Fig. 8 TimeGAN architecture. Extracted from [55]](https://figures.cochat.ai/paper-figures/f6de06e3be1a4ccf8d9f725384bb8b45/data-augmentation-techniques-in-time-series-domain-a-survey-and-taxonomy-figure-8.png)

![Fig. 9 Conditional SigWasserstein GAN training. Extracted from [111]](https://figures.cochat.ai/paper-figures/fb5a286212fe4c17ba536fe0586f5cdd/data-augmentation-techniques-in-time-series-domain-a-survey-and-taxonomy-figure-9.png)
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Data augmentation is one of the effective methods to generate augmented views in SSCL [47], [193].”
From Self-Supervised Learning for Time Series Analysis: Taxonomy, Progress, and Prospects (2024) · cites this paper as [47]
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474 · cited 2×
“In particular, they evaluated VGG[91], Residual Network[39], Multilayer Perceptron[100], Long Short-Term”
MethodVery Deep Convolutional Networks for Large-Scale Image Recognition
Simonyan, K., & Zisserman, A. (2014) · arXiv · doi:10.48550/arxiv.1409.1556 · Cited by 114,148 · cited 3×
“In particular, they evaluated VGG[91], Residual Network[39], Multilayer Perceptron[100], Long Short-Term”
MethodHochreiter, S., & Schmidhuber, J. (1997) · Neural Computation · doi:10.1162/neco.1997.9.8.1735 · Cited by 102,318 · cited 2×
“This conditional VAE (CVAE) architecture is composed of LSTM layers [92] which process the temporal data of the electrocardiogram (ECG) signals.”
BackgroundMethodGradient-based learning applied to document recognition
Lecun, Y., Bottou, L., Bengio, Y., & Haffner, P. (1998) · Proceedings of the IEEE · doi:10.1109/5.726791 · Cited by 58,805 · cited 1×
“Temporal Convolutional Neural Networks (CNNs) are CNNs[105] where the convolutional operation is calculated in 1D instead of traditional 2D convolution.”
MethodDensely Connected Convolutional Networks
Huang, G., Liu, Z., Van Der Maaten, L., & Weinberger, K. Q. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.243 · Cited by 48,566
Going deeper with convolutions
Szegedy, C., Liu, W., Jia, Y., Sermanet, P., Reed, S., Anguelov, D., Erhan, D., Vanhoucke, V., & Rabinovich, A. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7298594 · Cited by 48,562
Unpaired image-to-image translation using cycle-consistent adversarial networks
Zhu, J.-Y., Park, T., Isola, P., & Efros, A. A. (2017) · 2017 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2017.244 · Cited by 23,602
Image-to-Image Translation with Conditional Adversarial Networks
Isola, P., Zhu, J.-Y., Zhou, T., & Efros, A. A. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.632 · Cited by 22,930 · cited 1×
“This approximation was followed in GANs architectures such as [34, 48, 97, 103, 106], where a comparison between networks is possible using the same loss function to evaluate their training.”
MethodAdvances in neural information processing systems 7
Cios, K., & Shields, M. (1997) · Neurocomputing · doi:10.1016/s0925-2312(97)90018-7 · Cited by 22,059
On Information and Sufficiency
Kullback, S., & Leibler, R. A. (1951) · The Annals of Mathematical Statistics · doi:10.1214/aoms/1177729694 · Cited by 20,193 · cited 4×
“Recall that Kullback– Leibler divergence (KL divergence) [31] is defined as”
BackgroundMethodInfluentialGenerative adversarial networks
Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A., & Bengio, Y. (2020) · Communications of the ACM · doi:10.1145/3422622 · Cited by 18,384
Unsupervised representation learning with deep convolutional generative adversarial networks
Radford, A., Metz, L., & Chintala, S. (2015) · arXiv · doi:10.48550/arxiv.1511.06434 · Cited by 12,235 · cited 2×
“The former is in charge of generating the synthetic samples of the data distribution, while the latter tries to distinguish the real samples from the synthesised samples.”
BackgroundA Style-Based Generator Architecture for Generative Adversarial Networks
Karras, T., Laine, S., & Aila, T. (2019) · 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2019.00453 · Cited by 11,714 · cited 1×
“In addition to being still an open issue, there is consensus on some metrics and many papers measure their results with the same metrics[44, 45, 46, 47, 48, 49].”
BackgroundBidirectional recurrent neural networks
Schuster, M., & Paliwal, K. K. (1997) · IEEE Transactions on Signal Processing · doi:10.1109/78.650093 · Cited by 10,172 · cited 2×
“residual network [25], multilayer perceptron [33], long short-term memory [34], bidirectional long short-term memory [35] and long short-term memory fully convolutional network [36].”
BackgroundMethodConditional Generative Adversarial Nets
Mirza, M., & Osindero, S. (2014) · arXiv · doi:10.48550/arxiv.1411.1784 · Cited by 8,863 · cited 6×
“C-RNN-GAN [58] Biosignal DA based on GAN [95] LSTM-GAN Algorithm for Time Series Anomaly Detection [96] Intermittent Fault Data Generation Method Based on LSTM and GAN [97] GAN convolutional based GANomaly for Imbalanced Industrial Time Series [98] SpecGAN [61] TSDIGAN [99] T-CGAN [99] EmotionalGAN [99] WaveGAN [99] 1…”
InfluentialLearning long-term dependencies with gradient descent is difficult
Bengio, Y., Simard, P., & Frasconi, P. (1994) · IEEE Transactions on Neural Networks · doi:10.1109/72.279181 · Cited by 8,542
A Real-Time QRS Detection Algorithm
Pan, J., & Tompkins, W. J. (1985) · IEEE Transactions on Biomedical Engineering · doi:10.1109/tbme.1985.325532 · Cited by 7,855 · cited 1×
“This algorithm first detects features in the data, called characteristic points denoted with q, this is made by using the Pan-Tompkins QRS detector[77].”
MethodDynamic Programming Algorithm Optimization for Spoken Word Recognition
Sakoe, H., & Chiba, S. (1978) · IEEE Transactions on Acoustics Speech and Signal Processing · doi:10.1109/tassp.1978.1163055 · Cited by 6,727 · cited 9×
“DTW[87] is a classical algorithm that measures the similarity between two data sequences.”
BackgroundMethodInfluentialProgressive Growing of GANs for Improved Quality, Stability, and Variation
Karras, T., Aila, T., Laine, S., & Lehtinen, J. (2017) · arXiv · doi:10.48550/arxiv.1710.10196 · Cited by 4,727 · cited 1×
“The former is in charge of generating the synthetic samples of the data distribution, while the latter tries to distinguish the real samples from the synthesised samples.”
BackgroundSpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognition
Park, D. S., Chan, W., Zhang, Y., Chiu, C.-C., Zoph, B., Cubuk, E. D., & Le, Q. V. (2019) · Interspeech 2019 · doi:10.21437/interspeech.2019-2680 · Cited by 3,558 · cited 1×
“This algorithm has been used in several works, such as [52, 79].”
MethodNonlinear principal component analysis using autoassociative neural networks
Kramer, M. A. (1991) · AIChE Journal · doi:10.1002/aic.690370209 · Cited by 3,037
Andrzejak, R. G., Lehnertz, K., Mormann, F., Rieke, C., David, P., & Elger, C. E. (2001) · Physical Review E · doi:10.1103/physreve.64.061907 · Cited by 3,032
The Vanishing Gradient Problem During Learning Recurrent Neural Nets and Problem Solutions
Hochreiter, S. (1998) · International Journal of Uncertainty, Fuzziness and Knowledge-Based Systems · doi:10.1142/s0218488598000094 · Cited by 2,939 · cited 2×
“The internal composition of the discriminator is based on the work of Horchreiter[42] and Bengio et al.”
MethodTime series classification from scratch with deep neural networks: A strong baseline
Wang, Z., Yan, W., & Oates, T. (2017) · 2017 International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn.2017.7966039 · Cited by 2,299
DualGAN: Unsupervised Dual Learning for Image-to-Image Translation
Yi, Z., Zhang, H., Tan, P., & Gong, M. (2017) · 2017 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2017.310 · Cited by 2,134 · cited 1×
“This approximation was followed in GANs architectures such as [34, 48, 97, 103, 106], where a comparison between networks is possible using the same loss function to evaluate their training.”
MethodKowalska, M., Woźniak, M., Kijek, M., Mitrosz, P., Szakiel, J., & Turek, P. (2022) · Scientific Reports · doi:10.1038/s41598-021-99269-x · Cited by 2,117
Time-series forecasting with deep learning: a survey
Lim, B., & Zohren, S. (2021) · Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences · doi:10.1098/rsta.2020.0209 · Cited by 1,787 · cited 1×
“Previous works were centred on a specific model or approximation, with works such as [16, 21, 22].”
MethodData augmentation for improving deep learning in image classification problem
Mikolajczyk, A., & Grochowski, M. (2018) · 2018 International Interdisciplinary PhD Workshop (IIPhDW) · doi:10.1109/iiphdw.2018.8388338 · Cited by 1,697 · cited 1×
“Traditionally, in fields such as image recognition, different transformations have been applied to data such as cropping [23, 24, 25, 26], scaling [24, 25], mirroring [23, 27, 26], colour augmentation [23, 25, 28] or translation [27].”
MethodLSTM Fully Convolutional Networks for Time Series Classification
Karim, F., Majumdar, S., Darabi, H., & Chen, S. (2018) · IEEE Access · doi:10.1109/access.2017.2779939 · Cited by 1,465
Dau, H. A., Bagnall, A., Kamgar, K., Yeh, C.-C. M., Zhu, Y., Gharghabi, S., Ratanamahatana, C. A., & Keogh, E. (2019) · IEEE/CAA Journal of Automatica Sinica · doi:10.1109/jas.2019.1911747 · Cited by 1,128
Audio augmentation for speech recognition
Ko, T., Peddinti, V., Povey, D., & Khudanpur, S. (2015) · Interspeech 2015 · doi:10.21437/interspeech.2015-711 · Cited by 1,104
Pros and cons of GAN evaluation measures
Borji, A. (2019) · Computer Vision and Image Understanding · doi:10.1016/j.cviu.2018.10.009 · Cited by 963
A review of medical image data augmentation techniques for deep learning applications
Chlap, P., Min, H., Vandenberg, N., Dowling, J., Holloway, L., & Haworth, A. (2021) · Journal of Medical Imaging and Radiation Oncology · doi:10.1111/1754-9485.13261 · Cited by 956 · cited 1×
“In the latest times, several high-quality data augmentation review papers have been published [12, 13, 14].”
BackgroundSola, J., & Sevilla, J. (1997) · IEEE Transactions on Nuclear Science · doi:10.1109/23.589532 · Cited by 856 · cited 1×
“neural network (ANN) would be inefficient [115, 116] Although VAEs are algorithms capable of generating synthetic data, as opposed to traditional algorithms that only modify the original data, new neural network (NN) models such as GANs have mitigated their use in the field because by nature they are capable of genera…”
BackgroundUm, T. T., Pfister, F. M. J., Pichler, D., Endo, S., Lang, M., Hirche, S., Fietzek, U., & Kulić, D. (2017) · Proceedings of the 19th ACM International Conference on Multimodal Interaction · doi:10.1145/3136755.3136817 · Cited by 729 · cited 6×
“This technique was used in [70] where the data was captured from a wearable device to detect if a patient suffers from Parkinson’s disease.”
BackgroundMethodInfluentialTime Series Data Augmentation for Deep Learning: A Survey
Wen, Q., Sun, L., Yang, F., Song, X., Gao, J., Wang, X., & Xu, H. (2021) · Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence · doi:10.24963/ijcai.2021/631 · Cited by 727 · cited 1×
“Previous works were centred on a specific model or approximation, with works such as [21, 22, 16].”
MethodTime Series Data Augmentation for Deep Learning: A Survey
Wen, Q., Sun, L., Yang, F., Song, X., Gao, J., Wang, X., & Xu, H. (2020) · arXiv · doi:10.48550/arxiv.2002.12478 · Cited by 727
An empirical survey of data augmentation for time series classification with neural networks
Iwana, B. K., & Uchida, S. (2021) · PLOS ONE · doi:10.1371/journal.pone.0254841 · Cited by 725
A Survey of Data Augmentation Approaches for NLP
Feng, S., Gangal, V., Wei, J., Chandar, S., Vosoughi, S., Mitamura, T., & Hovy, E. (2021) · Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021 · doi:10.18653/v1/2021.findings-acl.84 · Cited by 670
Um, T. T. (n.d.) · mediaTUM – the media and publications repository of the Technical University Munich (Technical University Munich) · Cited by 602
DM-GAN: Dynamic Memory Generative Adversarial Networks for Text-To-Image Synthesis
Zhu, M., Pan, P., Chen, W., & Yang, Y. (2019) · 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2019.00595 · Cited by 590 · cited 1×
“In addition to being still an open issue, there is consensus on some metrics and many papers measure their results with the same metrics[44, 45, 46, 47, 48, 49].”
BackgroundData augmentation for deep-learning-based electroencephalography
Lashgari, E., Liang, D., & Maoz, U. (2020) · Journal of Neuroscience Methods · doi:10.1016/j.jneumeth.2020.108885 · Cited by 398
C-RNN-GAN: Continuous recurrent neural networks with adversarial training
Mogren, O. (2016) · arXiv · doi:10.48550/arxiv.1611.09904 · Cited by 386 · cited 6×
“Then they compared the distribution of the data in the two-dimensional space for TimeGAN[55], Recurrent Conditional GAN (GAN)[56], Continuous recurrent GAN (GAN)[58], T-Forcing[59], WaveNet[60] and WaveGAN[61].”
BackgroundMethodInfluentialData Augmentation for Deep Neural Network Acoustic Modeling
Cui, X., Goel, V., & Kingsbury, B. (2015) · IEEE/ACM Transactions on Audio, Speech, and Language Processing · doi:10.1109/taslp.2015.2438544 · Cited by 376
Data Augmentation for Brain-Tumor Segmentation: A Review
Nalepa, J., Marcinkiewicz, M., & Kawulok, M. (2019) · Frontiers in Computational Neuroscience · doi:10.3389/fncom.2019.00083 · Cited by 340 · cited 1×
“Furthermore, data generation algorithms produce new data so aggressively that much of the generated data is not possible, degenerating the quality of the augmented dataset [125].”
BackgroundTowards Principled Methods for Training Generative Adversarial Networks
Arjovsky, M., & Bottou, L. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1701.04862 · Cited by 335 · cited 1×
“Due to the GAN instability problems[121, 119, 122], most of the time a great portion of the samples synthesised by the G network lack of quality in certain aspects as the emergence of image artefacts [123].”
BackgroundConditional Sig-Wasserstein GANs for Time Series Generation
Liao, S., Ni, H., Sabate‐Vidales, M., Szpruch, Ł., Wiese, M., & Xiao, B. (2020) · Mathematical Finance · doi:10.1111/mafi.12423 · Cited by 293
DeLiGAN: Generative Adversarial Networks for Diverse and Limited Data
Gurumurthy, S., Sarvadevabhatla, R. K., & Babu, R. V. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.525 · Cited by 277 · cited 1×
“In these cases, traditional GAN metrics, such as the Inception Score[88], Mode Score[36] or Fréchet Inception Distance[40] are used to evaluate the results.”
MethodReal-valued (Medical) Time Series Generation with Recurrent Conditional GANs
Esteban, C., Hyland, S. L., & Rätsch, G. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1706.02633 · Cited by 253 · cited 3×
“Then they compared the distribution of the data in the two-dimensional space for TimeGAN[55], Recurrent Conditional GAN (GAN)[56], Continuous recurrent GAN (GAN)[58], T-Forcing[59], WaveNet[60] and WaveGAN[61].”
MethodAnomaly Detection of Time Series With Smoothness-Inducing Sequential Variational Auto-Encoder
Li, L., Yan, J., Wang, H., & Jin, Y. (2021) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2020.2980749 · Cited by 214 · cited 2×
“Another architecture based on the anomaly detection problem is the Smoothness-Inducing Sequential VAE (VAE)[88] which uses a VAE with recurrent layers to maintain temporal dependencies.”
MethodGANomaly: Semi-supervised Anomaly Detection via Adversarial Training
Akcay, S., Atapour-Abarghouei, A., & Breckon, T. P. (2019) · Lecture Notes in Computer Science · doi:10.1007/978-3-030-20893-6_39 · Cited by 188
A GAN-Based Anomaly Detection Approach for Imbalanced Industrial Time Series
Jiang, W., Hong, Y., Zhou, B., He, X., & Cheng, C. (2019) · IEEE Access · doi:10.1109/access.2019.2944689 · Cited by 186 · cited 3×
“…[96] Intermittent Fault Data Generation Method Based on LSTM and GAN [97] GAN convolutional based GANomaly for Imbalanced Industrial Time Series [98] SpecGAN [61] TSDIGAN [99] T-CGAN [99] EmotionalGAN [99] WaveGAN [99] 1D DCGAN for Fault Diagnosis of Electrical Machines [98] LSTMGAN [99]…”
BackgroundMethodInfluentialGenerating Synthetic Time Series to Augment Sparse Datasets
Forestier, G., Petitjean, F., Dau, H. A., Webb, G. I., & Keogh, E. (2017) · 2017 IEEE International Conference on Data Mining (ICDM) · doi:10.1109/icdm.2017.106 · Cited by 183
A Review: Generative Adversarial Networks
Gonog, L., & Zhou, Y. (2019) · 2019 14th IEEE Conference on Industrial Electronics and Applications (ICIEA) · doi:10.1109/iciea.2019.8833686 · Cited by 178 · cited 1×
“Due to the GAN instability problems[121, 119, 122], most of the time a great portion of the samples synthesised by the G network lack of quality in certain aspects as the emergence of image artefacts [123].”
BackgroundA survey on GANs for computer vision: Recent research, analysis and taxonomy
Iglesias, G., Talavera, E., & Díaz-Álvarez, A. (2023) · Computer Science Review · doi:10.1016/j.cosrev.2023.100553 · Cited by 173 · cited 1×
“This correction is used in the Gaussian latent distribution, where a shift hyperparameter λ is applied which is manually set to center the latent distribution, thus correcting the possible bias produced by missing values.”
MethodSelf-citationImproving the accuracy of global forecasting models using time series data augmentation
Bandara, K., Hewamalage, H., Liu, Y.-H., Kang, Y., & Bergmeir, C. (2021) · Pattern Recognition · doi:10.1016/j.patcog.2021.108148 · Cited by 169
Bringing a GAN to a Knife-Fight: Adapting Malware Communication to Avoid Detection
Rigaki, M., & Garcia, S. (2018) · 2018 IEEE Security and Privacy Workshops (SPW) · doi:10.1109/spw.2018.00019 · Cited by 167
Hsu, W.-N., Zhang, Y., & Glass, J. R. (2017) · IEEE Workshop on Automatic Speech Recognition and Understanding (ASRU) · doi:10.1109/asru.2017.8268911 · Cited by 158
Raghunathan, T. E. (2020) · Annual Review of Statistics and Its Application · doi:10.1146/annurev-statistics-040720-031848 · Cited by 158 · cited 1×
“Furthermore, in [20] they set out a further study of the repercussions of these technologies, highlighting one of the major advantages of generating synthesised data, the abstraction of privacy issues, and the ease of obtaining datasets.”
BackgroundHigh-Quality Facial Photo-Sketch Synthesis Using Multi-Adversarial Networks
Wang, L., Sindagi, V., & Patel, V. (2018) · 2018 13th IEEE International Conference on Automatic Face & Gesture Recognition (FG 2018) · doi:10.1109/fg.2018.00022 · Cited by 153
A review: preprocessing techniques and data augmentation for sentiment analysis
Duong, H.-T., & Nguyen-Thi, T.-A. (2021) · Computational Social Networks · doi:10.1186/s40649-020-00080-x · Cited by 146 · cited 1×
“Typically, these problems are addressed by preprocessing dataset techniques, such as subsampling, or in datasets that are not large enough, by Data Augmentation (DA) techniques [1, 2].”
MethodA survey on face data augmentation for the training of deep neural networks
Wang, X., Wang, K., & Lian, S. (2020) · Neural Computing and Applications · doi:10.1007/s00521-020-04748-3 · Cited by 141 · cited 1×
“Some authors [124] tend to differentiate between DA and data generation due to the great advances made in NN models.”
BackgroundDEVDAN: Deep evolving denoising autoencoder
Ashfahani, A., Pratama, M., Lughofer, E. D., & Ong, Y.-S. (2019) · Neurocomputing · doi:10.1016/j.neucom.2019.07.106 · Cited by 134
Systematic literature review of preprocessing techniques for imbalanced data
Felix, E. A., & Lee, S. P. (2019) · IET Software · doi:10.1049/iet-sen.2018.5193 · Cited by 132 · cited 1×
“Typically, these problems are addressed by pre-processing dataset techniques, such as subsampling, or in datasets that are not large enough, by Data Augmentation (DA) techniques [1, 2].”
MethodGenerative Adversarial Networks (GANs) in networking: A comprehensive survey & evaluation
Navidan, H., Moshiri, P. F., Nabati, M., Shahbazian, R., Ghorashi, S. A., Shah-Mansouri, V., & Windridge, D. (2021) · Computer Networks · doi:10.1016/j.comnet.2021.108149 · Cited by 131 · cited 1×
“There are many time series applications where these algorithms have shown good results when improving the capabilities of the models by enlarging the datasets [6, 7].”
BackgroundDemir, S., Mincev, K., Kok, K., & Paterakis, N. G. (2021) · Applied Energy · doi:10.1016/j.apenergy.2021.117695 · Cited by 130 · cited 1×
“In [87], they point out that most data augmentation methods for time series use feature space transformations to artificially enlarge the training set; they propose a composition of autoencoders ( AEs), variational autoencoders ( VAEs) and Wasserstein generative adversarial networks with gradient penalty (WGAN-GPs) fo…”
BackgroundGenerative Adversarial Networks for Spatio-temporal Data: A Survey
Gao, N., Xue, H., Shao, W., Zhao, S., Qin, K. K., Prabowo, A., Rahaman, M. S., & Salim, F. D. (2022) · ACM Transactions on Intelligent Systems and Technology · doi:10.1145/3474838 · Cited by 127
Conditional Sig-Wasserstein GANs for Time Series GenerationPreprint
Ni, H., Szpruch, Ł., Wiese, M., Liao, S., & Xiao, B. (2020) · SSRN Electronic Journal · doi:10.2139/ssrn.3623086 · Cited by 113
Biosignal Data Augmentation Based on Generative Adversarial Networks
Haradal, S., Hayashi, H., & Uchida, S. (2018) · 2018 40th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) · doi:10.1109/embc.2018.8512396 · Cited by 106 · cited 2×
“The work presented by Haradal et al.[95] also proposes a GAN architecture based on the implementation of LSTM cells in both the generator and discriminator networks to adapt to time series data.”
BackgroundMethodEEG-GAN: Generative adversarial networks for electroencephalograhic (EEG) brain signals
Hartmann, K. G., Schirrmeister, R. T., & Tonio Ball, A. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1806.01875 · Cited by 99 · cited 1×
“The model G generates samples trying to replicate, without copying, the distribution, while the model D discriminates the real samples from the fake samples.”
BackgroundDonahue, C., McAuley, J., & Puckette, M. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1802.04208 · Cited by 87 · cited 8×
“Then they compared the distribution of the data in the two-dimensional space for TimeGAN[55], Recurrent Conditional GAN (GAN)[56], Continuous recurrent GAN (GAN)[58], T-Forcing[59], WaveNet[60] and WaveGAN[61].”
BackgroundMethodInfluentialEnhancing human activity recognition using deep learning and time series augmented data
Alawneh, L. M., Alsarhan, T., Al-Zinati, M., Al‐Ayyoub, M., Jararweh, Y., & Lu, H. (2021) · Journal of Ambient Intelligence and Humanized Computing · doi:10.1007/s12652-020-02865-4 · Cited by 84 · cited 2×
“VAE for classification Enhancing human activity regonition using time series augmented data [82]”
MethodThe Wasserstein-Fourier Distance for Stationary Time Series
Cazelles, E., Robert, A., & Tobar, F. A. (2020) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2020.3046227 · Cited by 80 · cited 1×
“In [65] a novel measurement is proposed to quantify the distance between the time series distribution.”
BackgroundRamponi, G., Protopapas, P., Brambilla, M., & Janssen, R. R. P. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1811.08295 · Cited by 79 · cited 2×
“The Temporal-Conditional GAN (GAN)[106] is a GAN architecture based on the idea of transforming the Conditional GAN (GAN)[107] architecture to time series domain by replacing the 2D convolutional layers with the 1D convolutional layers.”
BackgroundInfluentialConvergence Problems with Generative Adversarial Networks (GANs)Preprint
Barnett, S. A. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1806.11382 · Cited by 77 · cited 1×
“The main problems that these networks suffer are Mode Collapse [117, 118], instability [119], convergence evaluation [120] and evaluation metrics.”
BackgroundA Novel LSTM-GAN Algorithm for Time Series Anomaly Detection
Zhu, G., Zhao, H., Liu, H., & Hua, S. (2019) · Prognostics and System Health Management Conference · doi:10.1109/phm-qingdao46334.2019.8942842 · Cited by 74 · cited 4×
“There are also hybrid implementations that combine 1D convolutions with other techniques, such as in [96] where LSTM-GAN is proposed.”
BackgroundMethodInfluentialData Augmentation with Suboptimal Warping for Time-Series Classification
Kamycki, K., Kapuscinski, T., & Oszust, M. (2019) · Sensors · doi:10.3390/s20010098 · Cited by 73
Huang, T., Chakraborty, P., & Sharma, A. (2021) · International Journal of Transportation Science and Technology · doi:10.1016/j.ijtst.2021.10.007 · Cited by 72
Naaz, F., Herle, A., Channegowda, J., Raj, A., & Lakshminarayanan, M. (2021) · International Journal of Energy Research · doi:10.1002/er.7013 · Cited by 62 · cited 1×
“This approach was followed in [57] where they applied tdistributed stochastic neighbour embedding (t-SNE) and principal component analysis (PCA).”
MethodRashid, K. M., & Louis, J. (2019) · International Symposium on Automation and Robotics in Construction (ISARC) · doi:10.22260/isarc2019/0087 · Cited by 60
A Survey of Synthetic Data Generation for Machine Learning
Abufadda, M., & Mansour, K. (2021) · International Arab Conference on Information Technology (ACIT) · doi:10.1109/acit53391.2021.9677302 · Cited by 52 · cited 1×
“Therefore, some studies have decided to take the path of building synthetic traffic generators to generate their datasets almost from scratch; some examples focus on this aspect [1, 33, 80].”
BackgroundGenerative adversarial networks in time series: A survey and taxonomyPreprint
Brophy, E., Wang, Z., She, Q., & Ward, T. E. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2107.11098 · Cited by 50 · cited 2×
“Previous works were centred on a specific model or approximation, with works such as [16, 21, 22].”
BackgroundMethodEmotionalGAN: Generating ECG to Enhance Emotion State Classification
Chen, G., Zhu, Y., Hong, Z., & Yang, Z. (2019) · International Conference on Artificial Intelligence and Computer Science (AICS) · doi:10.1145/3349341.3349422 · Cited by 50
Data Augmentation for Deep Learning-Based ECG Analysis
Pan, Q., Li, X., & Fang, L. (2020) · Springer eBooks · doi:10.1007/978-981-15-3824-7_6 · Cited by 48
Data Augmentation for Short-Term Time Series Prediction with Deep Learning
Flores, A., Tito-Chura, H., & Alanoca, H. A. (2021) · Lecture notes in networks and systems · doi:10.1007/978-3-030-80126-7_36 · Cited by 47
Deep time series models for scarce data
Wang, Q., Farahat, A., Gupta, C., & Zheng, S. (2021) · Neurocomputing · doi:10.1016/j.neucom.2020.12.132 · Cited by 39
Sensor-data augmentation for human activity recognition with time-warping and data masking
Jeong, C. Y., Shin, H. C., & Kim, M. (2021) · Multimedia Tools and Applications · doi:10.1007/s11042-021-10600-0 · Cited by 38 · cited 1×
“This algorithm has been used in several works, such as [52, 79].”
MethodON THE TRADEOFF BETWEEN MODE COLLAPSE AND SAMPLE QUALITY IN GENERATIVE ADVERSARIAL NETWORKS
Adiga, S., Attia, M. A., Chang, W.-T., & Tandon, R. (2018) · IEEE Global Conference on Signal and Information Processing · doi:10.1109/globalsip.2018.8646478 · Cited by 36 · cited 1×
“The main problems that these networks suffer are Mode Collapse [117, 118], instability [119], convergence evaluation [120] and evaluation metrics.”
BackgroundNazi, Z. A., Biswas, A., Khan, M. A. R., & Abir, T. A. (2019) · International Conference on Computer and Information Technology (ICCIT) · doi:10.1109/iccit48885.2019.9038287 · Cited by 32
On Mode Collapse in Generative Adversarial Networks
Zhang, K. (2021) · Lecture notes in computer science · doi:10.1007/978-3-030-86340-1_45 · Cited by 32 · cited 1×
“The main problems that these networks suffer are Mode Collapse [117, 118], instability [119], convergence evaluation [120] and evaluation metrics.”
BackgroundMulti-Perspective Discriminators-Based Generative Adversarial Network for Image Super Resolution
Lee, O.-Y., Shin, Y.-H., & Kim, J. (2019) · IEEE Access · doi:10.1109/access.2019.2942779 · Cited by 25
Fu, B., Kirchbuchner, F., & Kuijper, A. (2020) · International Conference on PErvasive Technologies Related to Assistive Environments (PETRA) · doi:10.1145/3389189.3392606 · Cited by 25 · cited 2×
“That is, in [30], they applied this algorithm by flipping the position of the sensors that recorded the data signals.”
MethodFeng, Q., Li, Y., & Wang, H. (2020) · Geophysics · doi:10.1190/geo2019-0815.1 · Cited by 23 · cited 2×
“Intelligent random noise modeling by improved VAE [83]”
BackgroundMethodSabir, R., Rosato, D., Hartmann, S., & Gühmann, C. (2021) · Proceedings - International Conference on Pattern Recognition/Proceedings/International Conference on Pattern Recognition · doi:10.1109/icpr48806.2021.9413119 · Cited by 23 · cited 1×
“This approximation has also been followed by Sabir et al.[109] for augmenting DC current signals, using samples of current signal with a frequency of 100 Hz during 16 seconds.”
MethodUsing Variational Autoencoder to augment Sparse Time series Datasets
Goubeaud, M., Jousen, P., Gmyrek, N., Ghorban, F., Schelkes, L., & Kummert, A. (2021) · International Conference on Optimization and Applications (ICOA) · doi:10.1109/icoa51614.2021.9442619 · Cited by 22 · cited 2×
“But the use of VAEs for DA is not only intended for neural network models, but can also improve results when traditional Machine Learning algorithms are applied [85].”
MethodSurvey: Image Mixing and Deleting for Data AugmentationPreprint
Naveed, H., Anwar, S., Hayat, M., et al. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2106.07085 · Cited by 18 · cited 1×
“Recently, a number of high-quality data augmentation review articles have been published [16, 27, 75].”
BackgroundPART-GAN: Privacy-Preserving Time-Series Sharing
Wang, S., Rudolph, C., Nepal, ., Grobler, M. M., & Chen, S. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-61609-0_46 · Cited by 17 · cited 1×
“Over the past few years, some works have suggested applying DA to time series data and treating it as if it were an image[50, 51].”
BackgroundVocal tract length perturbation and its application to male-female vocal tract shape conversion
Adachi, S., Takemoto, H., Kitamura, T., Mokhtari, P., & Honda, K. (2007) · The Journal of the Acoustical Society of America · doi:10.1121/1.2730743 · Cited by 16 · cited 1×
“Frequency warping is a variation of magnitude warping, mostly applied in speech processing[71, 72, 73].”
BackgroundGlowImp: Combining GLOW and GAN for Multivariate Time Series Imputation
Liu, C., Zhou, H., Sun, Z. Y., & Cui, G. (2022) · Lecture notes in computer science · doi:10.1007/978-3-030-95384-3_4 · Cited by 12 · cited 2×
“The GlowImp architecture[89] was proposed as a combination of the Wasserstein GAN (GAN)[93] architecture together with a VAE to impute missing data.”
MethodImproving the quality of generative models through Smirnov transformation
González-Prieto, Á., Mozó, A., Gómez-Canaval, S., & Talavera, E. (2022) · Information Sciences · doi:10.1016/j.ins.2022.07.066 · Cited by 11 · cited 1×
“Typical problems such as modal collapse, Nash equilibria, gradient vanishing or instability are suffered in every training of these models, making their optimisation a very complex process [127, 128].”
BackgroundSelf-citationCao, L., Horn, S., von Ehrenheim, V., Stahl, R. A., & Landgren, H. (2022) · ACM International Conference on Information and Knowledge Management (CIKM) · doi:10.1145/3511808.3557110 · Cited by 11
Survey of synthetic traffic generators
Patil, A., Surve, A., Gupta, A. K., Sharma, A., & Anmulwar, S. (2016) · International Conference on Inventive Computation Technologies (ICICT) · doi:10.1109/inventive.2016.7823282 · Cited by 10 · cited 1×
“Therefore, some studies have decided to take the path of building synthetic traffic generators to generate their datasets almost from scratch; some examples focus on this aspect [1, 33, 80].”
BackgroundAn Intermittent Fault Data Generation Method Based on LSTM and GAN
Shi, J., Ding, Y., & Lv, Z. (2021) · Global Reliability and Prognostics and Health Management Conference (PHM) · doi:10.1109/phm-nanjing52125.2021.9613001 · Cited by 6 · cited 2×
“The work presented by Shi et al.[97] uses the GAN architecture to generate sequences of faulty data from two different types.”
MethodOptimal Hyperparameters for Deep LSTM-Networks for Sequence Labeling Tasks
Reimers, N., & Gurevych, I. (2017) · arXiv · doi:10.48550/arxiv.1707.06799 · cited 1×
“Memory[85], Bidirectional Long Short-Term Memory[89] and Long Short-Term Memory Fully Convolutional Network[54].”
Research works printed but not linked: 23
- Goodfellow IJ, Pouget-Abadie J, Mirza M, Xu B, Warde-Farley D, Ozair S, Courville A, Bengio Y (2014) Generative adver- sarial networks 4. Lecun Y (1987) PhD Thesis: Modeles connexionnistes de L'apprentissage (connectionist Learning Models). Universite P. et M. Curie (Paris 6)
- Lecun Y (1987) PhD Thesis: Modeles connexionnistes de L’apprentissage (connectionist Learning Models). Universite P. et M. Curie (Paris 6)
- Bhowmik AK, Cabral P (2015) Spatially shifting temporal points: estimating pooled within-time series variograms for scarce hydrological data. Hydrol Earth Syst Sci Discuss 12(2):2243-2265
- Yang C-HH, Tsai Y-Y, Chen P-Y (2021) Voice2series: Repro- gramming acoustic models for time series classification. In: International conference on machine learning, pp. 11808-11819. PMLR
- Wang Z, Yan W, Oates T (2017) Time series classification from scratch with deep neural networks: a strong baseline. In: 2017 International joint conference on neural networks (IJCNN), pp. 1578-1585. IEEE · DOI · Paper page
- Karras T, Aittala M, Laine S, Ha ¨rko ¨nen E, Hellsten J, Lehtinen J, Aila T (2021) Alias-free generative adversarial networks. Adv Neural Inf Process Syst 34:852-863
- Gao L, Chen D, Zhao Z, Shao J, Shen HT (2021) Lightweight dynamic conditional gan with pyramid attention for text-to-im- age synthesis. Pattern Recognit 110:107384
- Heusel M, Ramsauer H, Unterthiner T, Nessler B, Hochreiter S (2017) Gans trained by a two time-scale update rule converge to a local nash equilibrium. Advances in neural information pro- cessing systems. In: Neural information processing systems (NIPS). 30
- Yoon J, Jarrett D, Van der Schaar M (2019) Time-series gen- erative adversarial networks. In: Advances in neural information processing systems 32
- Sutskever I, Martens J, Hinton GE (2011) Generating text with recurrent neural networks. In: ICML
- Van Den Oord A, Dieleman S, Zen H, Simonyan K, Vinyals O, Graves A, Kalchbrenner N, Senior AW, Kavukcuoglu K (2016) Wavenet: a generative model for raw audio. SSW 125:2
- Fu B, Kirchbuchner F, Kuijper A (2020) Data augmentation for time series: traditional vs generative models on capacitive proximity time series. In: Proceedings of the 13th ACM inter- national conference on pervasive technologies related to assis- tive environments, pp. 1-10 · DOI
- Debnath A, Waghmare G, Wadhwa H, Asthana S, Arora A (2021) Exploring generative data augmentation in multivariate time series forecasting: opportunities and challenges. Solar- Energy 137:52-560
- Cao P, Li X, Mao K, Lu F, Ning G, Fang L, Pan Q (2020) A novel data augmentation method to enhance deep neural net- works for detection of atrial fibrillation. Biomed Signal Process Control 56:101675
- Jaitly N, Hinton GE (2013) Vocal tract length perturbation (vtlp) improves speech recognition. In: Proceedings of ICML work- shop on deep learning for audio, speech and language, vol. 117
- Le Guennec A, Malinowski S, Tavenard R (2016) Data augmentation for time series classification using convolutional neural networks. In: ECML/PKDD workshop on advanced analytics and learning on temporal data
- Lee H, Hwang S, Shin J (2019) Rethinking data augmentation: self-supervision and self-distillation. arxiv 2019. arXiv preprint arXiv:1910.05872 · arXiv
- Moreno-Barea FJ, Jerez JM, Franco L (2020) Improving clas- sification accuracy using data augmentation on small data sets. Expert Syst Appl 161:113696
- Li J, Ren W, Han M (2021) Variational auto-encoders based on the shift correction for imputation of specific missing in multi- variate time series. Measurement 186:110055
- Arjovsky M, Chintala S, Bottou L (2017) Wasserstein genera- tive adversarial networks. In: International conference on machine learning, pp. 214-223. PMLR
- Higgins I, Matthey L, Pal A, Burgess C, Glorot X, Botvinick M, Mohamed S, Lerchner A (2016) beta-vae: learning basic visual concepts with a constrained variational framework. In Interna- tional conference on learning representations.
- Olszewski RT (2001) Generalized feature extraction for struc- tural pattern recognition in time -series data. PhD thesis 103.
- Shao J, Hu K, Wang C, Xue X, Raj B (2020) Is normalization indispensable for training deep neural network? Adv Neural Inf Process Syst 33:13434-13444 116.
Cited by
Self-Supervised Learning for Time Series Analysis: Taxonomy, Progress, and Prospects
Zhang, K., Wen, Q., Zhang, C., Cai, R., Jin, M., Liu, Y., Zhang, J. Y., Liang, Y., Pang, G., Song, D., et al. (2024) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2024.3387317 · Cited by 295
“Data augmentation is one of the effective methods to generate augmented views in SSCL [47], [193].”
BackgroundMethodA Comprehensive Survey on Data Augmentation
Wang, Z., Wang, P., Liu, K., Wang, P., Fu, Y., Lu, C.-T., Aggarwal, C. C., Pei, J., & Zhou, Y. (2025) · IEEE Transactions on Knowledge and Data Engineering · doi:10.1109/tkde.2025.3622600 · Cited by 281
“…by a certain function f θ , such that the augmented dataset ˜ D L = { ˜ X , ˜ y } is derived by: This representation also applies to an unlabeled dataset D U = { X } or a partially labeled dataset D P = { X L ∪ X U , y L } , but for simplicity, we only use a labeled dataset to explain the taxonomy.”
Avoiding common machine learning pitfalls
Lones, M. A. (2024) · Patterns · doi:10.1016/j.patter.2024.101046 · Cited by 170
Small data challenges for intelligent prognostics and health management: a review
Li, C., Li, S., Feng, Y., Gryllias, K., Gu, F., & PECHT, M. G. (2024) · Artificial Intelligence Review · doi:10.1007/s10462-024-10820-4 · Cited by 181
Strelcenia, E., & Prakoonwit, S. (2023) · Machine Learning and Knowledge Extraction · doi:10.3390/make5010019 · Cited by 127
“Furthermore, data augmentation regularizes the model and enhances its ability to generalize [73].”
Background
Metadata
- Authors
- Iglesias, G., Talavera, E., González-Prieto, Á., Mozo, A., & Gómez-Canaval, S.
- Affiliations
- Universidad Politécnica de Madrid (ES)
- Universidad Complutense de Madrid (ES)
- Corresponding author
- Talavera, Edgar
- Venue
- Neural Computing and Applications
- Published
- 2023-03-24
- Volume
- 35
- Issue
- 14
- Pages
- 10123-10145
- ISSN
- 0941-0643 (ISSN-L)
- 1433-3058
- OpenAlex
- W4360860566
- Cited by
- 347
- References
- 92
- Citations by year
- 2026: 86 · 2025: 143 · 2024: 94 · 2023: 21
- FWCI
- 41.69
- Type
- journal article
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- cc-by
- creativecommons.org/licenses/by/4.0 (tdm)
- creativecommons.org/licenses/by/4.0 (vor)
- Locations
- Neural Computing and Applications (publishedVersion, cc-by)
- Hispana (acceptedVersion, cc-by-nc-nd)
- arXiv (Cornell University) (submittedVersion)
- UPM Digital Archive (Technical University of Madrid) (submittedVersion, cc-by-nc-nd)
- arXiv (Cornell University)
- Funders
- Universidad Politécnica de Madrid
- Topics
- Physical Sciences › Computer Science › Signal Processing › Time Series Analysis and Forecasting
- Social Sciences › Decision Sciences › Management Science and Operations Research › Stock Market Forecasting Methods
- Physical Sciences › Computer Science › Artificial Intelligence › Neural Networks and Applications
- Keywords
- data augmentation, time series, synthetic data generation, deep generative models, noise injection, small datasets
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.