-
Automatic Classification of Galaxy Morphology: a rotationally invariant supervised machine learning method based on the UML-dataset
Authors:
G. W. Fang,
S. Ba,
Y. Z. Gu,
Z. S. Lin,
Y. J. Hou,
C. X. Qin,
C. C. Zhou,
J. Xu,
Y. Dai,
J. Song,
X. Kong
Abstract:
Classification of galaxy morphology is a challenging but meaningful task for the enormous amount of data produced by the next-generation telescope. By introducing the adaptive polar coordinate transformation, we develop a rotationally invariant supervised machine learning (SML) method that ensures consistent classifications when rotating galaxy images, which is always required to be satisfied phys…
▽ More
Classification of galaxy morphology is a challenging but meaningful task for the enormous amount of data produced by the next-generation telescope. By introducing the adaptive polar coordinate transformation, we develop a rotationally invariant supervised machine learning (SML) method that ensures consistent classifications when rotating galaxy images, which is always required to be satisfied physically but difficult to achieve algorithmically. The adaptive polar coordinate transformation, compared with the conventional method of data augmentation by including additional rotated images in the training set, is proved to be an effective and efficient method in improving the robustness of the SML methods. In the previous work, we generated a catalog of galaxies with well-classified morphologies via our developed unsupervised machine learning (UML) method. By using this UML-dataset as the training set, we apply the new method to classify galaxies into five categories (unclassifiable, irregulars, late-type disks, early-type disks, and spheroids). In general, the result of our morphological classifications following the sequence from irregulars to spheroids agrees well with the expected trends of other galaxy properties, including Sérsic indices, effective radii, nonparametric statistics, and colors. Thus, we demonstrate that the rotationally invariant SML method, together with the previously developed UML method, completes the entire task of automatic classification of galaxy morphology.
△ Less
Submitted 13 December, 2022;
originally announced December 2022.
-
Automatic morphological classification of galaxies: convolutional autoencoder and bagging-based multiclustering model
Authors:
C. C. Zhou,
Y. Z. Gu,
G. W. Fang,
Z. S. Lin
Abstract:
In order to obtain morphological information of unlabeled galaxies, we present an unsupervised machine-learning (UML) method for morphological classification of galaxies, which can be summarized as two aspects: (1) the methodology of convolutional autoencoder (CAE) is used to reduce the dimensions and extract features from the imaging data; (2) the bagging-based multiclustering model is proposed t…
▽ More
In order to obtain morphological information of unlabeled galaxies, we present an unsupervised machine-learning (UML) method for morphological classification of galaxies, which can be summarized as two aspects: (1) the methodology of convolutional autoencoder (CAE) is used to reduce the dimensions and extract features from the imaging data; (2) the bagging-based multiclustering model is proposed to obtain the classifications with high confidence at the cost of rejecting the disputed sources that are inconsistently voted. We apply this method on the sample of galaxies with $H<24.5$ in CANDELS. Galaxies are clustered into 100 groups, each contains galaxies with analogous characteristics. To explore the robustness of the morphological classifications, we merge 100 groups into five categories by visual verification, including spheroid, early-type disk, late-type disk, irregular, and unclassifiable. After eliminating the unclassifiable category and the sources with inconsistent voting, the purity of the remaining four subclasses are significantly improved. Massive galaxies ($M_*>10^{10}M_\odot$) are selected to investigate the connection with other physical properties. The classification scheme separates galaxies well in the U-V and V-J color space and Gini-$M_{20}$ space. The gradual tendency of Sérsic indexes and effective radii is shown from the spheroid subclass to the irregular subclass. It suggests that the combination of CAE and multi-clustering strategy is an effective method to cluster galaxies with similar features and can yield high-quality morphological classifications. Our study demonstrates the feasibility of UML in morphological analysis that would develop and serve the future observations made with China Space Station telescope.
△ Less
Submitted 29 December, 2021; v1 submitted 27 December, 2021;
originally announced December 2021.
-
A Distance-Deviation Consistency and Model-Independent Method to Test the Cosmic Distance-Duality Relation
Authors:
C. C. Zhou,
J. Hu,
M. C. LI,
X. Zhang,
G. W. Fang
Abstract:
A distance-deviation consistency and model-independent method to test the cosmic distance duality relation (CDDR) is provided. The method is worth attention on two aspects: firstly, a distance-deviation consistency method is used to pair subsamples: instead of pairing subsamples with redshift deviation smaller than a \textbf{value}, say $\left\vert Δz\right\vert <0.005$. The redshift deviation bet…
▽ More
A distance-deviation consistency and model-independent method to test the cosmic distance duality relation (CDDR) is provided. The method is worth attention on two aspects: firstly, a distance-deviation consistency method is used to pair subsamples: instead of pairing subsamples with redshift deviation smaller than a \textbf{value}, say $\left\vert Δz\right\vert <0.005$. The redshift deviation between subsamples decreases with the redshift to ensure the distance deviation stays the same. The method selects more subsamples at high redshift, up to $z=2.16$, and provides 120 subsample pairs. Secondly, the model-independent method involves the latest data set of $1048$ type Ia supernovae (SNe Ia) and $205$ strong gravitational lensing systems (SGLS), which are used to obtain the luminosity distances $D_L$ and the ratio of angular diameter distance $D_A$ respectively. With the model-independent method, parameters of the CDDR, the SNe Ia light-curve, and the SGLS are fitted simultaneously. \textbf{The result shows} that $η= 0.047^{+0.190}_{-0.151}$ and CDDR is validated at 1$σ$ confidence level for the form $\frac{D_L}{D_A}{(1 + z)^{ - 2}} =1+ ηz$.
△ Less
Submitted 13 November, 2020;
originally announced November 2020.
-
The Properties of Hα Emission-Line Galaxies at z = 2.24
Authors:
F. X. An,
X. Z. Zheng,
W. -H. Wang,
J. -S. Huang,
X. Kong,
J. -X. Wang,
G. W. Fang,
F. Zhu,
Q. -S. Gu,
H. Wu,
L. Hao,
X. -Y. Xia
Abstract:
Using deep narrow-band $H_2S1$ and $K_{s}$-band imaging data obtained with CFHT/WIRCam, we identify a sample of 56 H$α$ emission-line galaxies (ELGs) at $z=2.24$ with the 5$σ$ depths of $H_2S1=22.8$ and $K_{s}=24.8$ (AB) over 383 arcmin$^{2}$ area in the ECDFS. A detailed analysis is carried out with existing multi-wavelength data in this field. Three of the 56 H$α$ ELGs are detected in Chandra 4…
▽ More
Using deep narrow-band $H_2S1$ and $K_{s}$-band imaging data obtained with CFHT/WIRCam, we identify a sample of 56 H$α$ emission-line galaxies (ELGs) at $z=2.24$ with the 5$σ$ depths of $H_2S1=22.8$ and $K_{s}=24.8$ (AB) over 383 arcmin$^{2}$ area in the ECDFS. A detailed analysis is carried out with existing multi-wavelength data in this field. Three of the 56 H$α$ ELGs are detected in Chandra 4 Ms X-ray observation and two of them are classified as AGNs. The rest-frame UV and optical morphologies revealed by HST/ACS and WFC3 deep images show that nearly half of the H$α$ ELGs are either merging systems or with a close companion, indicating that the merging/interacting processes play a key role in regulating star formation at cosmic epoch z=2-3; About 14% are too faint to be resolved in the rest-frame UV morphology due to high dust extinction. We estimate dust extinction from SEDs. We find that dust extinction is generally correlated with H$α$ luminosity and stellar mass (SM). Our results suggest that H$α$ ELGs are representative of star-forming galaxies (SFGs). Applying extinction correction for individual objects, we examine the intrinsic H$α$ luminosity function (LF) at $z=2.24$, obtaining a best-fit Schechter function characterized by a faint-end slope of $α=-1.3$. This is shallower than the typical slope of $α\sim -1.6$ in previous works based on constant extinction correction. We demonstrate that this difference is mainly due to the different extinction corrections. The proper extinction correction is thus key to recovering the intrinsic LF as the extinction globally increases with H$α$ luminosity. Moreover, we find that our H$α$ LF mirrors the SM function of SFGs at the same cosmic epoch. This finding indeed reflects the tight correlation between SFR and SM for the SFGs, i.e., the so-called main sequence.
△ Less
Submitted 25 March, 2014; v1 submitted 14 February, 2014;
originally announced February 2014.
