Imbalance Learning for Variable Star Classification

The Astrophysical Journal, 2016

The need for the development of automatic tools to explore astronomical databases has been recognized since the inception of CCDs and modern computers. Astronomers already have developed solutions to tackle several science problems, such as automatic classification of stellar objects, outlier detection, and globular clusters identification, among others. New scientific problems emerge, and it is critical to be able to reuse the models learned before, without rebuilding everything from the beginning when the sciencientific problem changes. In this paper, we propose a new meta-model that automatically integrates existing classification models of variable stars. The proposed meta-model incorporates existing models that are trained in a different context, answering different questions and using different representations of data. A conventional mixture of expert algorithms in machine learning literature cannot be used since each expert (model) uses different inputs. We also consider the computational complexity of the model by using the most expensive models only when it is necessary. We test our model with EROS-2 and MACHO data sets, and we show that we solve most of the classification challenges only by training a meta-model to learn how to integrate the previous experts.

The Astrophysical Journal Supplement Series, 2012

With growing data volumes from synoptic surveys, astronomers necessarily must become more abstracted from the discovery and introspection processes. Given the scarcity of follow-up resources, there is a particularly sharp onus on the frameworks that replace these human roles to provide accurate and wellcalibrated probabilistic classification catalogs. Such catalogs inform the subsequent follow-up, allowing consumers to optimize the selection of specific sources for further study and permitting rigorous treatment of purities and efficiencies for population studies. Here, we describe a process to produce a probabilistic classification catalog of variability with machine learning from a multi-epoch photometric survey. In addition to producing accurate classifications, we show how to estimate calibrated class probabilities, and motivate the importance of probability calibration. We also introduce a methodology for feature-based anomaly detection, which allows discovery of objects in the survey that do not fit within the predefined class taxonomy. Finally, we apply these methods to sources observed by the All Sky Automated Survey (ASAS), and unveil the Machine-learned ASAS Classification Catalog (MACC), which is a 28-class probabilistic classification catalog of 50,124 ASAS sources. We estimate that MACC achieves a sub-20% classification error rate, and demonstrate that the class posterior probabilities are reasonably calibrated. MACC classifications compare favorably to the classifications of several previous domain-specific ASAS papers and to the

The Astrophysical Journal, 2011

With the coming data deluge from synoptic surveys, there is a growing need for frameworks that can quickly and automatically produce calibrated classification probabilities for newly-observed variables based on a small number of time-series measurements. In this paper, we introduce a methodology for variable-star classification, drawing from modern machine-learning techniques. We describe how to homogenize the information gleaned from light curves by selection and computation of real-numbered metrics (features), detail methods to robustly estimate periodic light-curve features, introduce treeensemble methods for accurate variable star classification, and show how to rigorously evaluate the classification results using cross validation. On a 25-class data set of 1542 well-studied variable stars, we achieve a 22.8% overall classification error using the random forest classifier; this represents a 24% improvement over the best previous classifier on these data. This methodology is effective for identifying samples of specific science classes: for pulsational variables used in Milky Way tomography we obtain a discovery efficiency of 98.2% and for eclipsing systems we find an efficiency of 99.1%, both at 95% purity. We show that the random forest (RF) classifier is superior to other machine-learned methods in terms of accuracy, speed, and relative immunity to features with no useful class information; the RF classifier can also be used to estimate the importance of each feature in classification. Additionally, we present the first astronomical use of hierarchical classification methods to incorporate a known class taxonomy in the classifier, which further reduces the catastrophic error rate to 7.8%. Excluding low-amplitude sources, our overall error rate improves to 14%, with a catastrophic error rate of 3.5%. 5 High-precision photometry missions (Kepler, MOST, CoRoT, etc.) are already challenging the theoretical understanding of the origin of variability and the connection of some specific sources to established classes of variables. 6 General Catalog of Variable Stars, http://www.sai.msu.su/groups/cluster/gcvs/gcvs/ 7 Not discussed herein are the challenges associated with discovery of variability. See Shin et al. (2009) for a review.

The Astrophysical Journal, 2012

Despite the great promise of machine-learning algorithms to classify and predict astrophysical parameters for the vast numbers of astrophysical sources and transients observed in large-scale surveys, the peculiarities of the training data often manifest as strongly biased predictions on the data of interest. Typically, training sets are derived from historical surveys of brighter, more nearby objects than those from more extensive, deeper surveys (testing data). This sample selection bias can cause catastrophic errors in predictions on the testing data because a) standard assumptions for machine-learned model selection procedures break down and b) dense regions of testing space might be completely devoid of training data. We explore possible remedies to sample selection bias, including importance weighting (IW), co-training (CT), and active learning (AL). We argue that AL-where the data whose inclusion in the training set would most improve predictions on the testing set are queried for manual follow-up-is an effective approach and is appropriate for many astronomical applications. For a variable star classification problem on a well-studied set of stars from Hipparcos and OGLE, AL is the optimal method in terms of error rate on the testing data, beating the off-the-shelf classifier by 3.4% and the other proposed methods by at least 3.0%. To aid with manual labeling of variable stars, we developed a web interface which allows for easy light curve visualization and querying of external databases. Finally, we apply active learning to classify variable stars in the ASAS survey, finding dramatic improvement in our agreement with the ACVS -2catalog, from 65.5% to 79.5%, and a significant increase in the classifier's average confidence for the testing set, from 14.6% to 42.9%, after a few AL iterations.

Monthly Notices of the Royal Astronomical Society, 2019

ABSTRACTIn the last years, automatic classification of variable stars has received substantial attention. Using machine learning techniques for this task has proven to be quite useful. Typically, machine learning classifiers used for this task require to have a fixed training set, and the training process is performed offline. Upcoming surveys such as the Large Synoptic Survey Telescope will generate new observations daily, where an automatic classification system able to create alerts online will be mandatory. A system with those characteristics must be able to update itself incrementally. Unfortunately, after training, most machine learning classifiers do not support the inclusion of new observations in light curves, they need to re-train from scratch. Naively re-training from scratch is not an option in streaming settings, mainly because of the expensive pre-processing routines required to obtain a vector representation of light curves (features) each time we include new observat...

Astronomy and Astrophysics, 2009

Context. The CoRoT space mission has two main scientific goals: exoplanet searches, and asteroseismology. Detecting planets using the occultation (or transit) method requires continuous monitoring of several thousand stars for a long period and with high photometric precision. As an important consequence, many high-quality light curves are obtained. Among this sample, a large fraction of variable stars is present, most of them previously unknown. This work describes the supervised classification of those newly measured variables, using automated methods. The methods were developed in the framework of the CoRoT mission, but they can easily be applied to other databases. Aims. In this work, we describe the pipeline for the fast supervised classification of light curves observed by the CoRoT exoplanet CCDs. We present the classification results obtained for the first four measured fields, which represent a one-year in-orbit operation. Methods. The basis of the adopted supervised classification methodology has been described in detail in a previous paper, as is its application to the OGLE database. Here, we present the modifications of the algorithms and of the training set, to optimize the performance when applied to the CoRoT data. Results. Classification results are presented for the observed fields IRa01, SRc01, LRc01, and LRa01 of the CoRoT mission. Statistics on the number of variables and the number of objects per class are given and typical light curves of high-probability candidates are shown. We also report on new stellar variability types discovered in the CoRoT data. The full classification results are publicly available.

Monthly Notices of the Royal Astronomical Society, 2017

During the past 10 years, a considerable amount of effort has been made to develop algorithms for automatic classification of variable stars. That has been primarily achieved by applying machine learning methods to photometric data sets where objects are represented as light curves. Classifiers require training sets to learn the underlying patterns that allow the separation among classes. Unfortunately, building training sets is an expensive process that demands a lot of human efforts. Every time data come from new surveys; the only available training instances are the ones that have a cross-match with previously labelled objects, consequently generating insufficient training sets compared with the large amounts of unlabelled sources. In this work, we present an algorithm that performs unsupervised classification of variable stars, relying only on the similarity among light curves. We tackle the unsupervised classification problem by proposing an untraditional approach. Instead of trying to match classes of stars with clusters found by a clustering algorithm, we propose a query-based method where astronomers can find groups of variable stars ranked by similarity. We also develop a fast similarity function specific for light curves, based on a novel data structure that allows scaling the search over the entire data set of unlabelled objects. Experiments show that our unsupervised model achieves high accuracy in the classification of different types of variable stars and that the proposed algorithm scales up to massive amounts of light curves.

arXiv: Astrophysics of Galaxies, 2019

We present a star/galaxy classification for the Southern Photometric Local Universe Survey (S-PLUS), based on a Machine Learning approach: the Random Forest algorithm. We train the algorithm using the S-PLUS optical photometry up to $r$=21, matched to SDSS/DR13, and morphological parameters. The metric of importance is defined as the relative decrease of the initial accuracy when all correlations related to a certain feature is vanished. In general, the broad photometric bands presented higher importance when compared to narrow ones. The influence of the morphological parameters has been evaluated training the RF with and without the inclusion of morphological parameters, presenting accuracy values of 95.0\% and 88.1\%, respectively. Particularly, the morphological parameter {\rm FWHM/PSF} performed the highest importance over all features to distinguish between stars and galaxies, indicating that it is crucial to classify objects into stars and galaxies. We investigate the misclass...

Lecture Notes in Computer Science, 2010

In this paper we present an experimental study of the performance of six machine learning algorithms applied to morphological galaxy classification. We also address the learning approach from imbalanced data sets, inherent to many real-world applications, such as astronomical data analysis problems. We used two over-sampling techniques: SMOTE and Resampling, and we vary the amount of generated instances for classification. Our experimental results show that the learning method Random Forest with Resampling obtain the best results for three, five and seven galaxy types, with a F-measure about .99 for all cases.

Machine learning has achieved an important role in the automatic classification of variable stars, and several classifiers have been proposed over the last decade. These classifiers have achieved impressive performance in several astronomical catalogues. However, some scientific articles have also shown that the training data therein contain multiple sources of bias. Hence, the performance of those classifiers on objects not belonging to the training data is uncertain, potentially resulting in the selection of incorrect models. Besides, it gives rise to the deployment of misleading classifiers. An example of the latter is the creation of open-source labelled catalogues with biased predictions. In this paper, we develop a method based on an informative marginal likelihood to evaluate variable star classifiers. We collect de-terministic rules that are based on physical descriptors of RR Lyrae stars, and then, to mitigate the biases, we introduce those rules into the marginal likelihood estimation. We perform experiments with a set of Bayesian Logistic Regressions, which are trained to classify RR Lyraes, and we found that our method outperforms traditional non-informative cross-validation strategies, even when penalized models are assessed. Our methodology provides a more rigorous alternative to assess machine learning models using astronomical knowledge. From this approach, applications to other classes of variable stars and algorithmic improvements can be developed.