Searching for Pulsars Using Image Pattern Recognition

Zhu, Weiwei; Berndsen, A.; Madsen, E. C.; Tan, Maxine; Stairs, I. H.; Brazier, A.; Lazarus, P.; Lynch, R.; Scholz, Paul; Stovall, K.; Ransom, S. M.; Banaszak, S.; Biwer, C.; Cohen, S L; Dartez, L. P.; Flanigan, J.; Lunsford, G.; Martínez, J.; Mata, A.; Rohr, M.; Walker, A.; Allen, B.; Bhat, N. D. R.; Bogdanov, Slavko; Camilo, F.; Chatterjee, Shami; Cordes, J. M.; Crawford, F.; Deneva, J. S.; Desvignes, G.; Ferdman, R. D.; Freire, P. C. C.; Hessels, J. W. T.; Jenet, F. A.; Kaplan, David L.; Kaspi, V. M.; Knispel, Benjamin; Lee, K. J.; Leeuwen, J. van; Lyne, A. G.; McLaughlin, M. A.; Siemens, X.; Spitler, Laura; Venkataraman, A.

doi:10.1088/0004-637x/781/2/117

Cited by 124 publications

(106 citation statements)

References 18 publications

Supporting

Mentioning

104

Contrasting

Unclassified

Order By: Relevance

“…As our focus is on machine learning, we only provide reviews of papers that use machine learning techniques Bates et al 2012;Zhu et al 2014;Morello et al 2014). The fact that these papers were all published in the last five years indicates that intelligent algorithms are becoming the new standard for pulsar classification.…”

Section: Related Work On Periodicity Searchesmentioning

confidence: 99%

“…Recently, Zhu et al (2014) created an artificial intelligence program to identify pulsars using image recognition algorithms called the Pulsar Image-based Classification System (PICS). PICS consists of two layers and was designed to emulate a human expert's visual identification process.…”

Section: Related Work On Periodicity Searchesmentioning

confidence: 99%

“…Artificial neural networks (ANNs) have been used in several related papers working with periodicity searches Bates et al 2012;Zhu et al 2014;Morello et al 2014), as mentioned in Section 3.2. The ANN we used is the Java implementation of a Multilayer Perceptron Network (MPN), which classifies instances using the supervised learning method of backpropagation and a sigmoid activation function in all neural nodes (Bishop 1995).…”

Section: Machine Learning Algorithmsmentioning

confidence: 99%

“…Automated approaches are the only feasible way to deal with big data, and offer many potential advantages to streamline the discovery process, e.g., by triggering the rapid follow up of candidates at multiple wavelengths to constrain their origins. Machine learning 1 algorithms have been applied to automatically detect pulsars in periodicity searches Bates et al 2012;Zhu et al 2014;Morello et al 2014). However, machine learning approaches with single-pulse candidates are lacking.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Detection of dispersed radio pulses: a machine learning approach to candidate identification and classification

Devine¹,

Goševa-Popstojanova²,

McLaughlin³

2016

Mon. Not. R. Astron. Soc.

View full text Add to dashboard Cite

Searching for extraterrestrial, transient signals in astronomical data sets is an active area of current research. However, machine learning techniques are lacking in the literature concerning single-pulse detection. This paper presents a new, two-stage approach for identifying and classifying dispersed pulse groups (DPGs) in single-pulse search output. The first stage identified DPGs and extracted features to characterize them using a new peak identification algorithm which tracks sloping tendencies around local maxima in plots of signal-to-noise ratio vs. dispersion measure. The second stage used supervised machine learning to classify DPGs. We created four benchmark data sets: one unbalanced and three balanced versions using three different imbalance treatments. We empirically evaluated 48 classifiers by training and testing binary and multiclass versions of six machine learning algorithms on each of the four benchmark versions. While each classifier had advantages and disadvantages, all classifiers with imbalance treatments had higher recall values than those with unbalanced data, regardless of the machine learning algorithm used. Based on the benchmarking results, we selected a subset of classifiers to classify the full, unlabelled data set of over 1.5 million DPGs identified in 42,405 observations made by the Green Bank Telescope. Overall, the classifiers using a multiclass ensemble tree learner in combination with two oversampling imbalance treatments were the most efficient; they identified additional known pulsars not in the benchmark data set and provided six potential discoveries, with significantly less false positives than the other classifiers.

show abstract

Section: Related Work On Periodicity Searchesmentioning

confidence: 99%

Section: Related Work On Periodicity Searchesmentioning

confidence: 99%

Section: Machine Learning Algorithmsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Detection of dispersed radio pulses: a machine learning approach to candidate identification and classification

Devine¹,

Goševa-Popstojanova²,

McLaughlin³

2016

Mon. Not. R. Astron. Soc.

View full text Add to dashboard Cite

show abstract

“…Candidates are uploaded to a web-based image-viewing and ranking application. 25 Periodicity candidates are analyzed with a pattern-recognition neural net (Zhu et al 2014), and single-pulse candidates are analyzed with a grouping algorithm (Karako-Argaman et al 2015;Chawla et al 2017 fast-folding algorithm for periodicity candidates (Parent et al in preparation) and a neural-net classifier for single-pulse candidates have also been implemented.…”

Section: Introductionmentioning

confidence: 99%

The Green Bank North Celestial Cap Pulsar Survey. III. 45 New Pulsar Timing Solutions

Lynch

Swiggum

Kondratiev

et al. 2018

ApJ

View full text Add to dashboard Cite

We provide timing solutions for 45 radio pulsars discovered by the Robert C. Byrd Green Bank Telescope. These pulsars were found in the Green Bank North Celestial Cap pulsar survey, an all-GBT-sky survey being carried out at a frequency of 350 MHz. We include pulsar timing data from the Green Bank Telescope and Low Frequency Array. Our sample includes five fully recycled millisecond pulsars (MSPs, three of which are in a binary system), a new relativistic double neutron star system, an intermediate-mass binary pulsar, a mode-changing pulsar, a 138 ms pulsar with a very low magnetic field, and several nulling pulsars. We have measured two post-Keplerian parameters and thus the masses of both objects in the double neutron star system. We also report a tentative companion mass measurement via Shapiro delay in a binary MSP. Two of the MSPs can be timed with high precision and have been included in pulsar timing arrays being used to search for low-frequency gravitational waves, while a third MSP is a member of the black widow class of binaries. Proper motion is measurable in five pulsars, and we provide an estimate of their space velocity. We report on an optical counterpart to a new black widow system and provide constraints on the optical counterparts to other binary MSPs. We also present a preliminary analysis of nulling pulsars in our sample. These results demonstrate the scientific return of long timing campaigns on pulsars of all types.

show abstract

Imbalanced Ensemble Learning for Enhanced Pulsar Identification

Holewik

Schaefer

Korovin

2020

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Pulsars can be detected based on their emitted radio waves. Machine learning methods can be employed to support automated screening of a large number of radio signals for pulsars. This is however a challenging task since training these methods is affected by an inherent imbalance in the acquired data with signals relating to actual pulsars being in the minority. In this paper, we demonstrate that ensemble classification methods that are dedicated to imbalanced classification problems can be successfully employed for pulsar identification. Classifier ensembles combine several individual classifiers to yield more robust and improved classification, while class imbalance can be addressed through careful sampling or through cost-sensitive classification. Experimental results, based on HTRU2 data, show that the investigated ensembles outperform methods that do not consider class balance, and suggest their use for other applications in astrophysics.

show abstract

Searching for Pulsars Using Image Pattern Recognition

Cited by 124 publications

References 18 publications

Detection of dispersed radio pulses: a machine learning approach to candidate identification and classification

Detection of dispersed radio pulses: a machine learning approach to candidate identification and classification

The Green Bank North Celestial Cap Pulsar Survey. III. 45 New Pulsar Timing Solutions

Imbalanced Ensemble Learning for Enhanced Pulsar Identification

Contact Info

Product

Resources

About