Star-Galaxy Classification in Multi-Band Optical Imaging

Beth Willman; David W. Hogg; Ross Fadely

RECORD

Star-Galaxy Classification in Multi-Band Optical Imaging

Ross Fadely, David W. Hogg, Beth Willman

Instrumentation and Methods for Astrophysics · Submitted 19 Jun 2012

abstract compiled · 2026-09-23 · no language-model review

The authors' main claim Thus, Hierarchical Bayesian template fitting may prove to be the optimal classification method in future surveys.

how it is stated · hedged possibility 1.00 read the record →

Comment (0) Read on arXiv ↗

Cite ▾

Claim paper

The compiled record

The paper's own sentences, arranged. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.

Open the compiled record

Compiled by classifier-1.13-20260917 via OpenRouter, question bank abstract-p1-20260923, 2026-09-23: 1 request, 9,590 input tokens, $0.0004 reported.

What the abstract itself states, as probabilities: claims first 0.03 · states limitation 0.93 · human participants 0.01 · controlled comparison 0.43 · randomized assignment 0.12 · states quantitative result 0.98 · compares with prior results 0.93 · mentions code or data availability 0.02.

The abstract, sentence by sentence

arXiv listing abstract v3. Results and the main claim are underlined; show every role.

Ground-based optical surveys such as PanSTARRS, DES, and LSST, will produce large catalogs to limiting magnitudes of r > 24. Star-galaxy separation poses a major challenge to such surveys because galaxies---even very compact galaxies---outnumber halo stars at these depths. We investigate photometric classification techniques on stars and galaxies with intrinsic FWHM < 0.2 arcsec. We consider unsupervised spectral energy distribution template fitting and supervised, data-driven Support Vector Machines (SVM). For template fitting, we use a Maximum Likelihood (ML) method and a new Hierarchical Bayesian (HB) method, which learns the prior distribution of template probabilities from the data. SVM requires training data to classify unknown sources; ML and HB don't. We consider i.) a best-case scenario (SVM_best) where the training data is (unrealistically) a random sampling of the data in both signal-to-noise and demographics, and ii.) a more realistic scenario where training is done on higher signal-to-noise data (SVM_real) at brighter apparent magnitudes. Testing with COSMOS ugriz data we find that HB outperforms ML, delivering ~80% completeness, with purity of ~60-90% for both stars and galaxies, respectively. We find no algorithm delivers perfect performance, and that studies of metal-poor main-sequence turnoff stars may be challenged by poor star-galaxy separation. Using the Receiver Operating Characteristic curve, we find a best-to-worst ranking of SVM_best, HB, ML, and SVM_real. We conclude, therefore, that a well trained SVM will outperform template-fitting methods. However, a normally trained SVM performs worse. Thus, Hierarchical Bayesian template fitting may prove to be the optimal classification method in future surveys.

background contribution method result interpretation main claim

What the classifier reports for each sentence. The number is the probability it gave the label.
#	sentence	role	the authors' own work	states the main conclusion	strength
s0	Ground-based optical surveys such as PanSTARRS, DES, and LSST, will produce large catalogs to limiting magnitudes of r > 24.	background 1.00	0.06	0.01	asserted 0.84
s1	Star-galaxy separation poses a major challenge to such surveys because galaxies---even very compact galaxies---outnumber halo stars at these depths.	background 1.00	0.07	0.01	asserted 1.00
s2	We investigate photometric classification techniques on stars and galaxies with intrinsic FWHM < 0.2 arcsec.	contribution 0.87	0.81	0.02	asserted 0.92
s3	We consider unsupervised spectral energy distribution template fitting and supervised, data-driven Support Vector Machines (SVM).	method 0.95	0.76	0.01	asserted 0.91
s4	For template fitting, we use a Maximum Likelihood (ML) method and a new Hierarchical Bayesian (HB) method, which learns the prior distribution of template probabilities from the data.	method 0.96	0.85	0.02	asserted 0.98
s5	SVM requires training data to classify unknown sources; ML and HB don't.	method 0.70	0.18	0.02	asserted 1.00
s6	We consider i.) a best-case scenario (SVM_best) where the training data is (unrealistically) a random sampling of the data in both signal-to-noise and demographics, and ii.) a more realistic scenario where training is done on higher signal-to-noise data (SVM_real) at brighter apparent magnitudes.	method 0.97	0.80	0.02	asserted 0.88
s7	Testing with COSMOS ugriz data we find that HB outperforms ML, delivering ~80% completeness, with purity of ~60-90% for both stars and galaxies, respectively.	result 1.00	0.97	0.11	asserted 1.00
s8	We find no algorithm delivers perfect performance, and that studies of metal-poor main-sequence turnoff stars may be challenged by poor star-galaxy separation.	result 0.54	0.95	0.06	asserted 0.57
s9	Using the Receiver Operating Characteristic curve, we find a best-to-worst ranking of SVM_best, HB, ML, and SVM_real.	result 0.99	0.96	0.06	asserted 1.00
s10	We conclude, therefore, that a well trained SVM will outperform template-fitting methods.	interpretation 0.79	0.49	0.50	asserted 1.00
s11	However, a normally trained SVM performs worse.	result 0.80	0.77	0.05	asserted 1.00
s12	Thus, Hierarchical Bayesian template fitting may prove to be the optimal classification method in future surveys.	interpretation 0.70	0.31	0.42	hedged possibility 1.00

What kind of paper the classifier reports this is: method or tool 0.52 · evidence from analysis of existing data 0.99.

What was examined, and what was not

sources read

arXiv listing abstract v3

not examined

full text
figures and tables
references and citing sentences
relations to other papers

levels run

P abstract

not run

S sections
T body sentences
C citations
R cross-paper relations

Every quoted sentence on this record is the authors'. Every label is from Pith's closed vocabulary, with the probability shown beside it. Every number in the measurements table is the authors', checked by code. Record as JSON

The paper

Original abstract

Ground-based optical surveys such as PanSTARRS, DES, and LSST, will produce large catalogs to limiting magnitudes of r > 24. Star-galaxy separation poses a major challenge to such surveys because galaxies---even very compact galaxies---outnumber halo stars at these depths. We investigate photometric classification techniques on stars and galaxies with intrinsic FWHM < 0.2 arcsec. We consider unsupervised spectral energy distribution template fitting and supervised, data-driven Support Vector Machines (SVM). For template fitting, we use a Maximum Likelihood (ML) method and a new Hierarchical Bayesian (HB) method, which learns the prior distribution of template probabilities from the data. SVM requires training data to classify unknown sources; ML and HB don't. We consider i.) a best-case scenario (SVM_best) where the training data is (unrealistically) a random sampling of the data in both signal-to-noise and demographics, and ii.) a more realistic scenario where training is done on higher signal-to-noise data (SVM_real) at brighter apparent magnitudes. Testing with COSMOS ugriz data we find that HB outperforms ML, delivering ~80% completeness, with purity of ~60-90% for both stars and galaxies, respectively. We find no algorithm delivers perfect performance, and that studies of metal-poor main-sequence turnoff stars may be challenged by poor star-galaxy separation. Using the Receiver Operating Characteristic curve, we find a best-to-worst ranking of SVM_best, HB, ML, and SVM_real. We conclude, therefore, that a well trained SVM will outperform template-fitting methods. However, a normally trained SVM performs worse. Thus, Hierarchical Bayesian template fitting may prove to be the optimal classification method in future surveys.

Human review

How Pith checks papers → · Full audit of this paper →

References

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

The relationship between morphology, density, and location in the cosmic web from massive to dwarf galaxies
astro-ph.GA 2026-09 accept novelty 7.0 of 10

Environmental segregation between early-type and late-type galaxies weakens with decreasing stellar mass and vanishes in the dwarf regime, where internal processes dominate galaxy evolution.
Impact of LSST systematics on stellar-stream density fluctuations for dark matter
astro-ph.GA 2026-09 accept novelty 7.0 of 10

Observational systematics from LSST degrade the sensitivity to dark matter subhalo-induced gaps in stellar streams by a factor of roughly 16 relative to idealized forecasts, with star-galaxy misclassification as the d...
Large Synoptic Survey Telescope: Dark Energy Science Collaboration
astro-ph.CO 2012-11 unverdicted novelty 2.0 of 10

The LSST Dark Energy Science Collaboration sets out its structure, its five cosmological probes, the systematics it considers most urgent, and a prioritized three-year work plan, without presenting new scientific results.

How the literature uses it

2 arXiv papers cite this in 3 sentences

Each citing sentence is taken from the citing paper's LaTeX source and typed by Pith's classifier. A label appears at probability 0.8 or more (0.95 for support), a contest at 0.7 or more; the quoted words are the citing authors'. How each label was checked · JSON

Papers citing it · 1 sentence without a confident label

Cites it as background 2 papers

Other supervised classifiers, such as Support Vectors Machine (SVM), have been more recently used for the star/galaxy separation problem, as well as unsupervised tools such as Hierarchical Bayesian techniques (e.g. [ref] ).

Star/galaxy separation at faint magnitudes: Application to a simulated Dark Energy Survey · Current tools for star-galaxy Separation · 0.90
This is explored in [ref] for SDSS Data Release 3 (DR3) and in [ref] for CFHTLenS and to select a pure star sample for Milky Way studies using SDSS DR7 in [ref] .

Effect of training characteristics on object classification: an application using Boosted Decision Trees · Introduction · 0.85

What it cites

Uses the method or tool of 1 paper

This marginalization is the total probability of a Gaussian distribution with variance $σ_C_i^2$—a value which is returned using least squares fitting techniques [ref] .

Data analysis recipes: Fitting a model to data · Template Fitting: Hierarchical Bayesian (HB) · 0.81

Cites as background 1 paper

Physically based Bayesian and $χ^2$ template fitting methods have been extensively used to infer the properties of galaxies [ref] .

PHAT: PHoto-z Accuracy Testing · Introduction · 0.89

Discussion

Know something about this paper? Leave a comment, or highlight any passage above to comment on it.