Generative adversarial networks in time series: A survey and taxonomy

Brophy, Eoin; Wang, Zhengwei; She, Qi; Ward, Tomas

doi:10.48550/arxiv.2107.11098

Cited by 16 publications

(22 citation statements)

References 47 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…While these successes have drawn much attention, GAN applications have diversified across disciplines such as time-series data generation. The work [3] gives a thorough summary of the GAN implementations in this field. The applicability of GANs to this type of data can solve many issues that current dataset holders face.…”

Section: Generative Adversarial Network (Gans)mentioning

confidence: 99%

See 1 more Smart Citation

TTS-GAN: A Transformer-based Time-Series Generative Adversarial Network

Li¹,

Metsis²,

Wang³

et al. 2022

Preprint

View full text Add to dashboard Cite

Signal measurements appearing in the form of time series are one of the most common types of data used in medical machine learning applications. However, such datasets are often small, making the training of deep neural network architectures ineffective. For time-series, the suite of data augmentation tricks we can use to expand the size of the dataset is limited by the need to maintain the basic properties of the signal. Data generated by a Generative Adversarial Network (GAN) can be utilized as another data augmentation tool. RNN-based GANs suffer from the fact that they cannot effectively model long sequences of data points with irregular temporal relations. To tackle these problems, we introduce TTS-GAN, a transformer-based GAN which can successfully generate realistic synthetic time-series data sequences of arbitrary length, similar to the real ones. Both the generator and discriminator networks of the GAN model are built using a pure transformer encoder architecture. We use visualizations and dimensionality reduction techniques to demonstrate the similarity of real and generated time-series data. We also compare the quality of our generated data with the best existing alternative, which is an RNN-based time-series GAN.

show abstract

Section: Generative Adversarial Network (Gans)mentioning

confidence: 99%

“…There has also been a movement towards using GANs for time series and sequential data generation, and forecasting. The review paper [3] gives a thorough summary of GAN implementations on time series data.…”

Section: Introductionmentioning

confidence: 99%

TTS-GAN: A Transformer-based Time-Series Generative Adversarial Network

Li¹,

Metsis²,

Wang³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…GANs can create new, synthetic data that mimic the real input by the application of multiple CNNs. GANs have been widely used for various applications, e.g., text and image translation [27], time series generation [28], and producing artificial images of Galaxies [29]. GANs use two CNNs, a generator and a discriminator, where the former is responsible for generating fake images from the input random noise, and the latter is used for discriminating the real and generated data, as shown in Figure 2.…”

Section: Generative Adversarial Networkmentioning

confidence: 99%

Generating transient noise artifacts in gravitational-wave detector data with generative adversarial networks

Powell¹,

Sun²,

Geréb³

et al. 2022

Preprint

View full text Add to dashboard Cite

Transient noise glitches in gravitational-wave detector data limit the sensitivity of searches and contaminate detected signals. In this Paper, we show how glitches can be simulated using generative adversarial networks. We produce hundreds of synthetic images for the 22 most common types of glitches seen in the LIGO, KAGRA, and Virgo detectors. The artificial glitches can be used to improve the performance of searches and parameter-estimation algorithms. We perform a neural network classification to show that our artificial glitches are an excellent match for real glitches, with an average classification accuracy across all 22 glitch types of 99.0%.

show abstract

“…This may, again, be used to generate class specific observations. While most generative adversarial network methods have been designed to work with visual data, methods which can be applied to time-series data have recently been proposed [12,143].…”

Section: Visualization Dimension Reduction and Semi-supervised Techni...mentioning

confidence: 99%

Fighting Money Laundering with Statistics and Machine Learning

Jensen¹,

Iosifidis²

2022

Preprint

View full text Add to dashboard Cite

Money laundering is a profound, global problem. Nonetheless, there is little statistical and machine learning research on the topic. In this paper, we focus on anti-money laundering in banks. To help organize existing research in the field, we propose a unifying terminology and provide a review of the literature. This is structured around two central tasks: (i) client risk profiling and (ii) suspicious behavior flagging. We find that client risk profiling is characterized by diagnostics, i.e., efforts to find and explain risk factors. Suspicious behavior flagging, on the other hand, is characterized by non-disclosed features and hand-crafted risk indices. Finally, we discuss directions for future research. One major challenge is the lack of public data sets. This may, potentially, be addressed by synthetic data generation. Other possible research directions include semi-supervised and deep learning, interpretability and fairness of the results.

show abstract

Generative adversarial networks in time series: A survey and taxonomy

Cited by 16 publications

References 47 publications

TTS-GAN: A Transformer-based Time-Series Generative Adversarial Network

TTS-GAN: A Transformer-based Time-Series Generative Adversarial Network

Generating transient noise artifacts in gravitational-wave detector data with generative adversarial networks

Fighting Money Laundering with Statistics and Machine Learning

Contact Info

Product

Resources

About