arXiv subjects

Artem Poliszczuk

Publications and source records attributed to Artem Poliszczuk.

25 records · Page 2Linked to original sources

arXiv2022

Using Multivariate Imputation by Chained Equations to Predict Redshifts of Active Galactic Nuclei

Redshift measurement of active galactic nuclei (AGNs) remains a time-consuming and challenging task, as it requires follow up spectroscopic observations and detailed analysis. Hence, there exists an urgent requirement for alternative redshift estimation techniques. The use of machine learning (ML) for this purpose has been growing over the last few years, primarily due to the availability of large-scale galactic surveys. However, due to observational errors, a significant fraction of these data sets often have missing entries, rendering that fraction unusable for ML regression applications. In this study, we demonstrate the performance of an imputation technique called Multivariate Imputation by Chained Equations (MICE), which rectifies the issue of missing data entries by imputing them using the available information in the catalog. We use the Fermi-LAT Fourth Data Release Catalog (4LAC) and impute 24% of the catalog. Subsequently, we follow the methodology described in Dainotti et al. (2021) and create an ML model for estimating the redshift of 4LAC AGNs. We present results which highlight positive impact of MICE imputation technique on the machine learning models performance and obtained redshift estimation accuracy.

astro-ph.IM↗

arXiv2022

Predicting the redshift of gamma-ray loud AGNs using Supervised Machine Learning: Part 2

Measuring the redshift of active galactic nuclei (AGNs) requires the use of time-consuming and expensive spectroscopic analysis. However, obtaining redshift measurements of AGNs is crucial as it can enable AGN population studies, provide insight into the star formation rate, the luminosity function, and the density rate evolution. Hence, there is a requirement for alternative redshift measurement techniques. In this project, we aim to use the Fermi gamma-ray space telescope's 4LAC Data Release (DR2) catalog to train a machine learning model capable of predicting the redshift reliably. In addition, this project aims at improving and extending with the new 4LAC Catalog the predictive capabilities of the machine learning (ML) methodology published in Dainotti et al. (2021). Furthermore, we implement feature engineering to expand the parameter space and a bias correction technique to our final results. This study uses additional machine learning techniques inside the ensemble method, the SuperLearner, previously used in Dainotti et al.(2021). Additionally, we also test a novel ML model called Sorted L-One Penalized Estimation (SLOPE). Using these methods we provide a catalog of estimated redshift values for those AGNs that do not have a spectroscopic redshift measurement. These estimates can serve as a redshift reference for the community to verify as updated Fermi catalogs are released with more redshift measurements.

astro-ph.HE↗

arXiv2021

Environmental Effects on AGN activity via Extinction-free Mid-Infrared Census

How does the environment affect active galactic nucleus (AGN) activity? We investigated this question in an extinction-free way, by selecting 1120 infrared galaxies in the $AKARI$ North Ecliptic Pole Wide field at redshift $z$ $\leq$ 1.2. A unique feature of the $AKARI$ satellite is its continuous 9-band infrared (IR) filter coverage, providing us with an unprecedentedly large sample of IR spectral energy distributions (SEDs) of galaxies. By taking advantage of this, for the first time, we explored the AGN activity derived from SED modelling as a function of redshift, luminosity, and environment. We quantified AGN activity in two ways: AGN contribution fraction (ratio of AGN luminosity to the total IR luminosity), and AGN number fraction (ratio of number of AGNs to the total galaxy sample). We found that galaxy environment (normalised local density) does not greatly affect either definitions of AGN activity of our IRG/LIRG samples (log ${\rm L}_{\rm TIR}$ $\leq$ 12). However, we found a different behavior for ULIRGs (log ${\rm L}_{\rm TIR}$ $>$ 12). At our highest redshift bin (0.7 $\lesssim$ z $\lesssim$ 1.2), AGN activity increases with denser environments, but at the intermediate redshift bin (0.3 $\lesssim$ z $\lesssim$ 0.7), the opposite is observed. These results may hint at a different physical mechanism for ULIRGs. The trends are not statistically significant (p $\geq$ 0.060 at the intermediate redshift bin, and p $\geq$ 0.139 at the highest redshift bin). Possible different behavior of ULIRGs is a key direction to explore further with future space missions (e.g., $JWST$, $Euclid$, $SPHEREx$).

astro-ph.GA↗

arXiv2021

Active galactic nuclei catalog from the AKARI NEP Wide field

Context. The North Ecliptic Pole (NEP) field provides a unique set of panchromatic data, well suited for active galactic nuclei (AGN) studies. Selection of AGN candidates is often based on mid-infrared (MIR) measurements. Such method, despite its effectiveness, strongly reduces a catalog volume due to the MIR detection condition. Modern machine learning techniques can solve this problem by finding similar selection criteria using only optical and near-infrared (NIR) data. Aims. Aims of this work were to create a reliable AGN candidates catalog from the NEP field using a combination of optical SUBARU/HSC and NIR AKARI/IRC data and, consequently, to develop an efficient alternative for the MIR-based AKARI/IRC selection technique. Methods. A set of supervised machine learning algorithms was tested in order to perform an efficient AGN selection. Best of the models were formed into a majority voting scheme, which used the most popular classification result to produce the final AGN catalog. Additional analysis of catalog properties was performed in form of the spectral energy distribution (SED) fitting via the CIGALE software. Results. The obtained catalog of 465 AGN candidates (out of 33 119 objects) is characterized by 73% purity and 64% completeness. This new classification shows consistency with the MIR-based selection. Moreover, 76% of the obtained catalog can be found only with the new method due to the lack of MIR detection for most of the new AGN candidates. Training data, codes and final catalog are available via the github repository. Final AGN candidates catalog will be also available via the CDS service after publication.

astro-ph.GA↗

arXiv2021

An Active Galactic Nucleus Recognition Model based on Deep Neural Network

To understand the cosmic accretion history of supermassive black holes, separating the radiation from active galactic nuclei (AGNs) and star-forming galaxies (SFGs) is critical. However, a reliable solution on photometrically recognising AGNs still remains unsolved. In this work, we present a novel AGN recognition method based on Deep Neural Network (Neural Net; NN). The main goals of this work are (i) to test if the AGN recognition problem in the North Ecliptic Pole Wide (NEPW) field could be solved by NN; (ii) to shows that NN exhibits an improvement in the performance compared with the traditional, standard spectral energy distribution (SED) fitting method in our testing samples; and (iii) to publicly release a reliable AGN/SFG catalogue to the astronomical community using the best available NEPW data, and propose a better method that helps future researchers plan an advanced NEPW database. Finally, according to our experimental result, the NN recognition accuracy is around 80.29% - 85.15%, with AGN completeness around 85.42% - 88.53% and SFG completeness around 81.17% - 85.09%.

astro-ph.GA↗

arXiv2019

AGN selection in the AKARI NEP deep field with the fuzzy SVM algorithm

The aim of this work is to create a new catalog of reliable AGN candidates selected from the AKARI NEP-Deep field. Selection of the AGN candidates was done by applying a fuzzy SVM algorithm, which allows to incorporate measurement uncertainties into the classification process. The training dataset was based on the spectroscopic data available for selected objects in the NEP-Deep and NEP-Wide fields. The generalization sample was based on the AKARI NEP-Deep field data including objects without optical counterparts and making use of the infrared information only. A high quality catalog of previously unclassified 275 AGN candidates was prepared.

astro-ph.IM↗

arXiv2017

Searching for previously unknown classes of objects in the AKARI-NEP Deep data with fuzzy logic SVM algorithm

In this proceedings application of a fuzzy Support Vector Machine (FSVM) learning algorithm, to classify mid-infrared (MIR) sources from the AKARI NEP Deep field into three classes: stars, galaxies and AGNs, is presented. FSVM is an improved version of the classical SVM algorithm, incorporating measurement errors into the classification process; this is the first successful application of this algorithm in the astronomy. We created reliable catalogues of galaxies, stars and AGNs consisting of objects with MIR measurements, some of them with no optical counterparts. Some examples of identified objects are shown, among them O-rich and C-rich AGB stars.

astro-ph.GA↗