English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Time-domain astronomy is progressing rapidly with the ongoing and upcominglarge-scale photometric sky surveys led by the Vera C. Rubin Observatoryproject (LSST). Billions of variable sources call for better automaticclassification algorithms for light curves. Among them, periodic variable starsare frequently studied. Different categories of periodic variable stars have ahigh degree of class imbalance and pose a challenge to algorithms includingdeep learning methods. We design two kinds of architectures of neural networksfor the classification of periodic variable stars in the Catalina Survey's DataRelease 2: a multi-input recurrent neural network (RNN) and a compound networkcombing the RNN and the convolutional neural network (CNN). To deal with classimbalance, we apply Gaussian Process to generate synthetic light curves withartificial uncertainties for data augmentation. For better performance, weorganize the augmentation and training process in a "bagging-like" ensemblelearning scheme. The experimental results show that the better approach is thecompound network combing RNN and CNN, which reaches the best result of 86.2% onthe overall balanced accuracy and 0.75 on the macro F1 score. We develop theensemble augmentation method to solve the data imbalance when classifyingvariable stars and prove the effectiveness of combining differentrepresentations of light curves in a single model. The proposed methods wouldhelp build better classification algorithms of periodic time series data forfuture sky surveys (e.g., LSST).

Periodic Variable Star Classification with Deep Learning: Handling Data Imbalance in an Ensemble Augmentation Way