Towards Synthetic Multivariate Time Series Generation for Flare Forecasting

Chen, Yang; Kempton, Dustin J.; Ahmadzadeh, Azim; Angryk, Rafal A.

doi:10.1007/978-3-030-87986-0_26

Cited by 13 publications

(9 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Works on multivariate time-series synthesis in general are related given our problem involves generating time-series data for multiple radio network KPIs. Existing work [10,30,31], however, targets very different problems from ours. For instance, in [30], an unconditional GAN based multivariate time-series synthesis model is introduced to generate data for resource utilization measurement of CDN caches whereas we target a conditional data generation problem.…”

Section: Background 21 Related Workmentioning

confidence: 98%

“…For instance, in [30], an unconditional GAN based multivariate time-series synthesis model is introduced to generate data for resource utilization measurement of CDN caches whereas we target a conditional data generation problem. As another example, Chen et al [10] focus on mitigating the severe class imbalance in the data for predicting rare events (e.g., solar flares). Among these works, DoppelGANger (DG) [31] is a more closely related work that is aimed at unconditional GAN based generation of multivariate time-series data for networks and systems (e.g., Wikipedia article views over time, network monitoring data over time, resource usage in compute clusters).…”

Section: Background 21 Related Workmentioning

confidence: 99%

See 1 more Smart Citation

GenDT

Sun

Marina

et al. 2022

Proceedings of the 18th International Conference on Emerging Networking EXperiments and Technologies

View full text Add to dashboard Cite

Drive testing continues to play a key role in mobile network optimization for operators but its high cost is a big concern. Alternative approaches like virtual drive testing (VDT) target device testing in the lab whereas MDT or crowdsourcing based approaches are limited by the incentives users have to participate and contribute measurements. With the aim of augmenting drive testing and significantly reducing its cost, we propose GenDT, a novel deep generative model that synthesizes high-fidelity time series of key radio network key performance indicators (KPIs). The training of GenDT relies on a relatively small amount of real-world measurement data along with corresponding and easily accessible network and environment context data. Through this, GenDT learns the relationship between context and radio network KPIs as they vary over time, and therefore trained GenDT model can subsequently be relied on to generate time series for different KPIs for new drive test routes (trajectories) without having to collect field measurements. GenDT represents an initial attempt at enabling efficient drive testing via generative modeling. Evaluations with real-world mobile network drive testing measurement datasets from two countries demonstrate that GenDT can synthesize significantly more dependable data than a range of baselines. We further show that GenDT has the potential to significantly reduce the drive testing related measurement effort, and that GenDT-generated data yields similar results to that with real data in the context of two downstream use cases -QoE prediction and handover analysis. CCS CONCEPTS• Networks → Mobile networks; • Computing methodologies → Neural networks.

show abstract

Section: Background 21 Related Workmentioning

confidence: 98%

Section: Background 21 Related Workmentioning

confidence: 99%

GenDT

Sun

Marina

et al. 2022

Proceedings of the 18th International Conference on Emerging Networking EXperiments and Technologies

View full text Add to dashboard Cite

show abstract

“…A thorough discussion on the methods used for data collection, cleaning, and pre-processing of the solar active region and flare data is published and the entire project is made open-sourced as well (SWAN-SF Project, 2020). Several studies investigated different applications of SWAN-SF, such as time series profiling (Ma et al, 2019), outlier detection from spatiotemporal data (Cai et al, 2020b), PIL detection (Cai et al, 2020a), synthetic MVTS generation (Chen et al, 2021), robust sampling in extreme class-imbalance cases , flare classification using Time Series Forest , SEP forecasting (Ji et al, in press), and MVTS feature ranking (Yeolekar et al, 2021). Additionally, a data challenge series was organized by DMLab on this benchmark data set, as part of the IEEE Big Data Conference, in 2019 and 2020 (IEEE Big Data Cup Challenge 2019, 2019; IEEE Big Data Cup Challenge 2020, 2020).…”

Section: Ml-ready Data Sets At Gsumentioning

confidence: 99%

Revisiting the Solar Research Cyberinfrastructure Needs: A White Paper of Findings and Recommendations

Nita¹,

Ahmadzadeh²,

Criscuoli³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Solar and Heliosphere physics are areas of remarkable data-driven discoveries. Recent advances in high-cadence, high-resolution multiwavelength observations, growing amounts of data from realistic modeling, and operational needs for uninterrupted science-quality data coverage generate the demand for a solar metadata standardization and overall healthy data infrastructure. This white paper is prepared as an effort of the working group "Uniform Semantics and Syntax of Solar Observations and Events" created within the "Towards Integration of Heliophysics Data, Modeling, and Analysis Tools" EarthCube Research Coordination Network (@HDMIEC RCN), with primary objectives to discuss current advances and identify future needs for the solar research cyberinfrastructure. The white paper summarizes presentations and discussions held during the special working group session at the EarthCube Annual Meeting on June 19th, 2020, as well as community contribution gathered during a series of preceding workshops and subsequent RCN working group sessions. The authors provide examples of the current standing of the solar research cyberinfrastructure, and describe the problems related to current data handling approaches. The list of the top-level recommendations agreed by the authors of the current white paper is presented at the beginning of the paper.

show abstract

“…There are several data augmentation techniques ranging from basic to advanced methods, such as statistical generative models, decomposition methods, and deep-learning methods (Wen et al 2020). Moreover, the synthetic multivariate time series generation for flare forecasting has been a successful method to overcome such rare events, as explained in Chen et al (2021).…”

Section: Introductionmentioning

confidence: 99%

Improving Solar Energetic Particle Event Prediction through Multivariate Time Series Data Augmentation

Hosseinzadeh,

Filali Boubrahimi,

Hamdi

2024

ApJS

View full text Add to dashboard Cite

Solar energetic particles (SEPs) are associated with extreme solar events that can cause major damage to space- and ground-based life and infrastructure. High-intensity SEP events, particularly ∼100 MeV SEP events, can pose severe health risks for astronauts owing to radiation exposure and affect Earth’s orbiting satellites (e.g., Landsat and the International Space Station). A major challenge in the SEP event prediction task is the lack of adequate SEP data because of the rarity of these events. In this work, we aim to improve the prediction of ∼30, ∼60, and ∼100 MeV SEP events by synthetically increasing the number of SEP samples. We explore the use of a univariate and multivariate time series of proton flux data as input to machine-learning-based prediction methods, such as time series forest (TSF). Our study covers solar cycles 22, 23, and 24. Our findings show that using data augmentation methods, such as the synthetic minority oversampling technique, remarkably increases the accuracy and F1-score of the classifiers used in this research, especially for TSF, where the average accuracy increased by 20%, reaching around 90% accuracy in the ∼100 MeV SEP prediction task. We also achieved higher prediction accuracy when using the multivariate time series data of the proton flux. Finally, we build a pipeline framework for our best-performing model, TSF, and provide a comprehensive hierarchical classification of the ∼100, ∼60, and ∼30 MeV and non-SEP prediction scenarios.

show abstract

Towards Synthetic Multivariate Time Series Generation for Flare Forecasting

Cited by 13 publications

References 19 publications

GenDT

GenDT

Revisiting the Solar Research Cyberinfrastructure Needs: A White Paper of Findings and Recommendations

Improving Solar Energetic Particle Event Prediction through Multivariate Time Series Data Augmentation

Contact Info

Product

Resources

About