اشترك بالحزمة الذهبية واحصل على وصول غير محدود شمرا أكاديميا
تسجيل مستخدم جديدDeep Neural Network Classifier for Variable Stars with Novelty Detection Capability
65
0
0.0
(
0
)
اسأل ChatGPT حول البحث
ﻻ يوجد ملخص باللغة العربية
Common variable star classifiers are built only with the goal of producing the correct class labels, leaving much of the multi-task capability of deep neural networks unexplored. We present a periodic light curve classifier that combines a recurrent neural network autoencoder for unsupervised feature extraction and a dual-purpose estimation network for supervised classification and novelty detection. The estimation network optimizes a Gaussian mixture model in the reduced-dimension feature space, where each Gaussian component corresponds to a variable class. An estimation network with a basic structure of a single hidden layer attains a cross-validation classification accuracy of ~99%, on par with the conventional workhorses, random forest classifiers. With the addition of photometric features, the network is capable of detecting previously unseen types of variability with precision 0.90, recall 0.96, and an F1 score of 0.93. The simultaneous training of the autoencoder and estimation network is found to be mutually beneficial, resulting in faster autoencoder convergence, and superior classification and novelty detection performance. The estimation network also delivers adequate results even when optimized with pre-trained autoencoder features, suggesting that it can readily extend existing classifiers to provide added novelty detection capabilities.
قيم البحث
اقرأ أيضاً
Astronomical surveys of celestial sources produce streams of noisy time series measuring flux versus time (light curves). Unlike in many other physical domains, however, large (and source-specific) temporal gaps in data arise naturally due to intrani
ght cadence choices as well as diurnal and seasonal constraints. With nightly observations of millions of variable stars and transients from upcoming surveys, efficient and accurate discovery and classification techniques on noisy, irregularly sampled data must be employed with minimal human-in-the-loop involvement. Machine learning for inference tasks on such data traditionally requires the laborious hand-coding of domain-specific numerical summaries of raw data (features). Here we present a novel unsupervised autoencoding recurrent neural network (RNN) that makes explicit use of sampling times and known heteroskedastic noise properties. When trained on optical variable star catalogs, this network produces supervised classification models that rival other best-in-class approaches. We find that autoencoded features learned on one time-domain survey perform nearly as well when applied to another survey. These networks can continue to learn from new unlabeled observations and may be used in other unsupervised tasks such as forecasting and anomaly detection.
With recent developments in imaging and computer technology the amount of available astronomical data has increased dramatically. Although most of these data sets are not dedicated to the study of variable stars much of it can, with the application o
f proper software tools, be recycled for the discovery of new variable stars. Fits Viewer and Data Retrieval System is a new software package that takes advantage of modern computer advances to search astronomical data for new variable stars. More than 200 new variable stars have been found in a data set taken with the Calvin College Rehoboth Robotic telescope using FVDRS. One particularly interesting example is a very fast subdwarf B with a 95 minute orbital period, the fastest currently known of the HW Vir type.
The Dark Matter Particle Explorer (DAMPE) is a space-borne particle detector and cosmic ray observatory in operation since 2015, designed to probe electrons and gamma rays from a few GeV to 10 TeV energy, as well as cosmic protons and nuclei up to 10
0 TeV. Among the main scientific objectives is the precise measurement of the cosmic electron+positron flux, which due to the very large proton background in orbit requires a powerful particle identification method. In the past decade, the field of machine learning has provided us the needed tools. This paper presents a neural network based approach to cosmic electron identification and proton rejection and showcases its performances based on simulated Monte Carlo data. The neural network reaches significantly lower background than the classical, cut-based method for the same detection efficiency, especially at highest energies. A good matching between simulations and real data completes the picture.
Photometric variability detection is often considered as a hypothesis testing problem: an object is variable if the null-hypothesis that its brightness is constant can be ruled out given the measurements and their uncertainties. Uncorrected systemati
c errors limit the practical applicability of this approach to high-amplitude variability and well-behaving data sets. Searching for a new variability detection technique that would be applicable to a wide range of variability types while being robust to outliers and underestimated measurement uncertainties, we propose to consider variability detection as a classification problem that can be approached with machine learning. We compare several classification algorithms: Logistic Regression (LR), Support Vector Machines (SVM), k-Nearest Neighbors (kNN) Neural Nets (NN), Random Forests (RF) and Stochastic Gradient Boosting classifier (SGB) applied to 18 features (variability indices) quantifying scatter and/or correlation between points in a light curve. We use a subset of OGLE-II Large Magellanic Cloud (LMC) photometry (30265 light curves) that was searched for variability using traditional methods (168 known variable objects identified) as the training set and then apply the NN to a new test set of 31798 OGLE-II LMC light curves. Among 205 candidates selected in the test set, 178 are real variables, 13 low-amplitude variables are new discoveries. We find that the considered machine learning classifiers are more efficient (they find more variables and less false candidates) compared to traditional techniques that consider individual variability indices or their linear combination. The NN, SGB, SVM and RF show a higher efficiency compared to LR and kNN.
Stars exhibit a bewildering variety of variable behaviors ranging from explosive magnetic flares to stochastically changing accretion to periodic pulsations or rotations. The principal LSST surveys will have cadences too sparse and irregular to captu
re most of these phenomena. A novel idea is proposed here to observe a single Galactic field, rich in unobscured stars, in a continuous sequence of $sim 15$ second exposures for one long winter night in a single photometric band. The result will be a unique dataset of $sim 1$ million regularly spaced stellar lightcurves. The lightcurves will gives a particularly comprehensive collection of dM star variability. A powerful array of statistical procedures can be applied to the ensemble of lightcurves from the long-standing fields of time series analysis, signal processing and econometrics. Dozens of `features describing the variability can be extracted and subject to machine learning classification, giving a unique authoritative objective classification of rapidly variable stars. The most effective features can then inform the wider LSST community on the best approaches to variable star identification and classification from the sparse, irregular cadences that dominate the LSST project.
سجل دخول لتتمكن من نشر تعليقات
التعليقات
جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
