arXiv subjects

Marek Sikora

Publications and source records attributed to Marek Sikora.

At least 19 recordsLinked to original sources

arXiv2025

RuleKit 2: Faster and simpler rule learning

Rules offer an invaluable combination of predictive and descriptive capabilities. Our package for rule-based data analysis, RuleKit, has proven its effectiveness in classification, regression, and survival problems. Here we present its second version. New algorithms and optimized implementations of those previously included, significantly improved the computational performance of our suite, reducing the analysis time of some data sets by two orders of magnitude. The usability of RuleKit 2 is provided by two new components: Python package and browser application with a graphical user interface. The former complies with scikit-learn, the most popular data mining library for Python, allowing RuleKit 2 to be straightforwardly integrated into existing data analysis pipelines. RuleKit 2 is available at GitHub under GNU AGPL 3 license (https://github.com/adaa-polsl/RuleKit)

cs.LG↗

arXiv2024

Towards consistency of rule-based explainer and black box model -- fusion of rule induction and XAI-based feature importance

Rule-based models offer a human-understandable representation, i.e. they are interpretable. For this reason, they are used to explain the decisions of non-interpretable complex models, referred to as black box models. The generation of such explanations involves the approximation of a black box model by a rule-based model. To date, however, it has not been investigated whether the rule-based model makes decisions in the same way as the black box model it approximates. Decision making in the same way is understood in this work as the consistency of decisions and the consistency of the most important attributes used for decision making. This study proposes a novel approach ensuring that the rule-based surrogate model mimics the performance of the black box model. The proposed solution performs an explanation fusion involving rule generation and taking into account the feature importance determined by the selected XAI methods for the black box model being explained. The result of the method can be both global and local rule-based explanations. The quality of the proposed solution was verified by extensive analysis on 30 tabular benchmark datasets representing classification problems. Evaluation included comparison with the reference method and an illustrative case study. In addition, the paper discusses the possible pathways for the application of the rule-based approach in XAI and how rule-based explanations, including the proposed method, meet the user perspective and requirements for both content and presentation. The software created and a detailed report containing the full experimental results are available on the GitHub repository (https://github.com/ruleminer/FI-rules4XAI ).

cs.AI↗

arXiv2023

Observations of a black hole X-ray binary indicate formation of a magnetically arrested disk

Accretion of material onto a black hole drags any magnetic fields present inwards, increasing their strength. Theory predicts that sufficiently strong magnetic fields can halt the accretion flow, producing a magnetically arrested disk (MAD). We analyze archival multi-wavelength observations of an outburst from the black hole X-ray binary MAXI J1820+070 in 2018. The radio and optical fluxes are delayed by about 8 and 17 days respectively, compared to the X-ray flux. We interpret this as evidence for the formation of a MAD. In this scenario, the magnetic field is amplified by an expanding corona, forming a MAD around the time of the radio peak. The optical delay is then due to thermal viscous instability in the outer disk.

astro-ph.HE↗

arXiv2023

Formation and evolution of transient jets and their cavities in black-hole X-ray binaries

We propose a model explaining the origin of transient/episodic jets in black-hole X-ray binaries, in which they are caused by transitions from a collimated, strongly magnetized, jet to a wide, un-collimated, outflow. The change occurs when the accretion flow leaves the magnetically-choked state due to an increase of the accretion rate for a weakly varying magnetic flux. The formed powerful jet then detaches from its base, and propagates as a discrete ejection. The uncollimated outflow then produces a relativistic plasma that fills surrounding of the black hole, contributing to the formation of a low-density cavity. While the pressure in the cavity is in equilibrium with the surrounding interstellar medium (ISM), its inertia is orders of magnitude lower than that of the ISM. This implies that the plasma cannot efficiently decelerate the ejecta, explaining most of the observations. The modest deceleration within the cavities observed in some cases can be then due to the presence of clouds and/or filaments, forming a wide transition zone between the cavity and the ISM.

astro-ph.HE↗

arXiv2023

No Need for an Extreme Jet Energy in the Black-Hole X-Ray Binary MAXI J1348--630

We model interaction with the surrounding medium of the main discrete jet ejection in the accreting black-hole binary MAXI J1348--630. The kinetic energy in the ejection of that jet was estimated before to be $>10^{46}$ erg. That energy requires that the jet power was about two orders of magnitude above the limit corresponding to a magnetically arrested accretion onto a maximally rotating black hole. That large estimate was obtained by considering the initial ballistic jet propagation in a surrounding cavity followed by a sudden deceleration in interstellar medium under the assumption of its standard density of $\sim$1 cm$^{-3}$. Such densities are likely in the surrounding of this source given its location in the Galactic Plane. Here, we show that the estimate of the kinetic energy can be reduced to realistic values of $\sim\! 10^{44}$ erg by considering the presence of a transition layer with an exponential density growth separating the cavity and the interstellar medium. In that case, the jet is found to decelerate mostly in the transition layer, in regions with the densities $\ll$1 cm$^{-3}$, which strongly reduces the energy requirement. Still, the required jet masses are large, ruling out the presence of a significant number of electron-positron pairs.

astro-ph.HE↗

arXiv2023

Separate and conquer heuristic allows robust mining of contrast sets in classification, regression, and survival data

Identifying differences between groups is one of the most important knowledge discovery problems. The procedure, also known as contrast sets mining, is applied in a wide range of areas like medicine, industry, or economics. In the paper we present RuleKit-CS, an algorithm for contrast set mining based on separate and conquer - a well established heuristic for decision rule induction. Multiple passes accompanied with an attribute penalization scheme provide contrast sets describing same examples with different attributes, distinguishing presented approach from the standard separate and conquer. The algorithm was also generalized for regression and survival data allowing identification of contrast sets whose label attribute/survival prognosis is consistent with the label/prognosis for the predefined contrast groups. This feature, not provided by the existing approaches, further extends the usability of RuleKit-CS. Experiments on over 130 data sets from various areas and detailed analysis of selected cases confirmed RuleKit-CS to be a useful tool for discovering differences between defined groups. The algorithm was implemented as a part of the RuleKit suite available at GitHub under GNU AGPL 3 licence (https://github.com/adaa-polsl/RuleKit). Keywords: contrast sets, separate and conquer, regression, survival

cs.DB↗

arXiv2022

A simple analytical model of magnetic jets

We propose a simple analytical jet model of magnetic jets, in which radially-averaged profiles of main physical quantities are obtained based on conservation laws and some results of published GRMHD jet simulations. We take into account conversion of the magnetic energy flux to bulk acceleration in jets formed around rotating black holes assuming the mass continuity equation and constant jet power, which leads to the Bernoulli equation. For assumed profiles of the bulk Lorentz factor and the radius, this gives us the profile of the toroidal magnetic field component along the jet. We then consider the case where the poloidal field component is connected to a rotating black hole surrounded by an accretion disc. Our formalism then recovers the standard formula for the power extracted from a rotating black hole. We find that the poloidal field strength dominates over the toroidal one in the comoving frame up to large distances, which means that jets should be more stable to current-driven kink modes. The resulting magnetic field profiles can then be used to calculate the jet synchrotron emission.

astro-ph.HE↗

arXiv2022

The Composition and Power of the Jet of the Broad-Line Radio Galaxy 3C 120

We calculated the electron-positron pair production rate at the base of the jet of 3C 120 by collisions of photons from the hot accretion flow using the measurement of its average soft gamma-ray spectrum by Compton Gamma Ray Observatory. We found that this rate approximately equals the flow rate of the leptons emitting the observed synchrotron radio-to-IR spectrum of the jet core, calculated using the extended jet model following Blandford \& Konigl. This coincidence shows the jet composition is likely to be pair-dominated. We then calculated the jet power in the bulk motion of ions, and found it greatly exceeds that achievable by the magnetically arrested disk scenario for the maximum black hole spin unless the jet contains mostly pairs. Next, we found that the magnetic flux through the synchrotron-emitting jet equals the maximum poloidal flux that can thread the black hole. Finally, we compared two estimates of the magnetization parameter at the onset of the synchrotron emission and found they are in agreement only if pairs dominate the jet content.

astro-ph.HE↗

arXiv2021

Jet Parameters in the Black-Hole X-Ray Binary MAXI J1820+070

We study the jet in the hard state of the accreting black-hole binary MAXI J1820+070. From the available radio-to-optical spectral and variability data, we put strong constraints on the jet parameters. We find while it is not possible to uniquely determine the jet Lorentz factor from the spectral and variability properties alone, we can estimate the jet opening angle ($\approx 1.5\pm1^\circ$), the distance at which the jet starts emitting synchrotron radiation ($\sim 3 \times10^{10}$cm), and the magnetic field strength there ($\sim$10$^4$G), with relatively low uncertainty, as they depend weakly on the bulk Lorentz factor. We find the breaks in the variability power spectra from radio to sub-mm are consistent with variability damping over the time scale equal to the travel time along the jet at any Lorentz factor. This factor can still be constrained by the electron-positron pair production rate within the jet base, which we calculate based on the observed X-ray/soft gamma-ray spectrum, and the jet power, required to be less than the accretion power. The minimum ($\sim$1.5) and maximum ($\sim$4.5) Lorentz factors correspond to the dominance of pairs and ions, and the minimum and maximum jet power, respectively. We estimate the magnetic flux threading the black hole and find the jet can be powered by the Blandford-Znajek mechanism in a magnetically-arrested flow accretion flow. We point out the similarity of our derived formalism to that of core shifts, observed in extragalactic radio sources.

astro-ph.HE↗

arXiv2021

The dependence of the fraction of radio luminous quasars on redshift and its theoretical implications

While radio emission in quasars can be contributed to by a variety of processes (involving star forming regions, accretion disk coronas and winds, and jets), the powering of the radio loudest quasars must involve very strong jets, presumably launched by the Blandford-Znajek mechanism incorporating the magnetically arrested disk (MAD) scenario. We focus on the latter and investigate the dependence of their fraction on redshift. We also examine the dependence of the radio-loud fraction (RLF) on BH mass ($M_{\rm BH}$) and Eddington ratio ($λ_{\rm Edd}$) while excluding the redshift bias by narrowing its range. In both these investigations we remove the bias associated with: (1) the diversity of source selection by constructing two well-defined, homogeneous samples of quasars (first within $0.7 \leq z < 1.9$, second within $0.5 \leq z < 0.7$); (2) a strong drop in the RLF of quasars at smaller BH masses by choosing those with BH masses larger than $10^{8.5} M_{\odot}$. We confirm some previous results showing the increase in the fraction of radio-loud quasars with cosmic time and that this trend can be even steeper if we account for the bias introduced by the dependence of the RLF on BH mass whereas the bias introduced by the dependence of the RLF on Eddington ratio is shown to be negligible. Assuming that quasar activities are triggered by galaxy mergers we argue that such an increase can result from the slower drop with cosmic time of mixed mergers than of wet mergers.

astro-ph.HE↗

arXiv2021

SCARI: Separate and Conquer Algorithm for Action Rules and Recommendations Induction

This article describes an action rule induction algorithm based on a sequential covering approach. Two variants of the algorithm are presented. The algorithm allows the action rule induction from a source and a target decision class point of view. The application of rule quality measures enables the induction of action rules that meet various quality criteria. The article also presents a method for recommendation induction. The recommendations indicate the actions to be taken to move a given test example, representing the source class, to the target one. The recommendation method is based on a set of induced action rules. The experimental part of the article presents the results of the algorithm operation on sixteen data sets. As a result of the conducted research the Ac-Rules package was made available.

cs.AI↗

arXiv2021

Orientation of the crescent image of M 87*

The first image of the black hole (BH) M 87* obtained by the Event Horizon Telescope (EHT) has the shape of a crescent extending from the E to WSW position angles, while the observed direction of the large-scale jet is WNW. Images based on numerical simulations of BH accretion flows suggest that on average the projected BH spin axis should be oriented SSW. We explore highly simplified toy models for geometric distribution and kinematics of emitting regions in the Kerr metric, perform ray tracing to calculate the corresponding images, and simulate their observation by the EHT to calculate the corresponding visibilities and closure phases. We strictly assume that (1) the BH spin vector is fixed to the jet axis, (2) the emitting regions are stationary and symmetric with respect to the BH spin, and that (3) the emissivities are isotropic in the local rest frames. Emission from the crescent sector between SSE and WSW can be readily explained in terms of an equatorial ring with either circular or plunging geodesic flows, regardless of the value of BH spin. In the case of plane-symmetric polar caps with plunging geodesic flows, the dominant image of the cap located behind the BH is sensitive to the angular momentum of the emitter. Within the constraints of our model, we have not found a viable explanation for the observed brightness of the ESE sector. Most likely, the ESE "hotspot" has been produced by a non-stationary localised perturbation in the inner accretion flow. Alternatively, it could result from locally anisotropic synchrotron emissivities. Multi-epoch and polarimetric results from the EHT will be essential to verify the theoretically expected alignment of the BH spin with the large-scale jet.

astro-ph.HE↗

arXiv2020

On the diversity of the jet production efficiency in Swift/BAT AGNs

We address the very large diversity of the jet production efficiency in active galactic nuclei (AGNs) by using data on low redshift AGNs selected from the Swift/BAT catalog and having black hole (BH) masses larger than $10^{8.5}\,M_{\odot}$. Most of these AGNs accrete at intermediate rates and have bolometric luminosities dominated by mid-IR radiation. Our sample contains $14\%$ radio-loud (RL), $6\%$ radio-intermediate (RI), and $80\%$ radio-quiet (RQ) AGNs. All RL objects are found to have extended radio structures and most of them have classical FR II morphology. Converting their radio loudness to the jet production efficiency, we find that the median of this efficiency is on the order of $(ε_d/0.1)\%$, where $ε_d=L_{\rm bol}/\dot{M}c^2$ is the radiation efficiency of the accretion disk. Without knowing the contribution of jets to the radio emission in the RQ AGNs, we are only able to estimate their efficiencies using upper limits. Their median is found to be $0.002(ε_d/0.1)\%$. Our results suggest that some threshold conditions must be satisfied to allow production of strong, relativistic jets in RL AGNs. We discuss several possible scenarios and argue that the production of collimated, relativistic jets must involve the Blandford-Znajek mechanism and can be activated only in those AGNs whose lifetime is longer than the time required to enter the magnetically arrested disk (MAD). Presuming that MAD is required to collimate relativistic jets, we expect that the weak nonrelativistic jets observed in some RQ AGNs are produced by accretion disks rather than by rotating BHs.

astro-ph.HE↗

arXiv2020

Identifying radio active galactic nuclei among radio-emitting galaxies

Basing our analysis on ROGUE I, a catalog of over 32,000 radio sources associated with optical galaxies, we provide two diagnostics to select the galaxies where the radio emission is due to an active galactic nucleus (AGN). Each of these diagnostics can be applied independently. The first one, dubbed MIRAD, compares the flux $F_{W3}$ in the $W3$ mid-infrared band of the WISE telescope, with the radio flux at 1.4 GHz, $\Frad$. MIRAD requires no optical spectra. The second diagnostic, dubbed DLM, relates the 4000 Å break strength, $D_{\rm n}(4000)$, with the radio luminosity per unit stellar mass. The DLM diagram has already been used in the past, but not as standalone. For these two diagrams, we propose simple, empirical dividing lines that result in the same classification for the objects in common. These lines correctly classify as radio-AGN 99.5 percent of the extended radio sources in the ROGUE~I catalog, and as star-forming (SF) galaxies 98--99 percent of the galaxies identified as such by their emission line ratios. Both diagrams clearly show that radio AGNs are preferentially found among elliptical galaxies and among galaxies hosting the most massive black holes. Most of the radio sources classified as radio-AGNs in the MIRAD or DLM diagrams are either optically weak AGNs or retired galaxies.

astro-ph.GA↗

arXiv2020

Classification supporting COVID-19 diagnostics based on patient survey data

Distinguishing COVID-19 from other flu-like illnesses can be difficult due to ambiguous symptoms and still an initial experience of doctors. Whereas, it is crucial to filter out those sick patients who do not need to be tested for SARS-CoV-2 infection, especially in the event of the overwhelming increase in disease. As a part of the presented research, logistic regression and XGBoost classifiers, that allow for effective screening of patients for COVID-19, were generated. Each of the methods was tuned to achieve an assumed acceptable threshold of negative predictive values during classification. Additionally, an explanation of the obtained classification models was presented. The explanation enables the users to understand what was the basis of the decision made by the model. The obtained classification models provided the basis for the DECODE service (decode.polsl.pl), which can serve as support in screening patients with COVID-19 disease. Moreover, the data set constituting the basis for the analyses performed is made available to the research community. This data set consisting of more than 3,000 examples is based on questionnaires collected at a hospital in Poland.

cs.LG↗

arXiv2020

On the significance of relativistically hot pairs in the jets of FR II radio galaxies

The energetic composition of radio lobes in the FR II galaxies $-$ estimated by comparing their radio luminosities with the powers required to inflate cavities in the external medium $-$ seems to exclude the possibility of their energetic domination by protons. Furthermore, if the jets were dominated by the kinetic energy of cold protons, it would be difficult to efficiently accelerate leptons in the jets' terminal shocks. Assuming that the relative energy contents of leptons, protons and magnetic fields are preserved across the shocks, the above implies that the large-scale jets should also be energetically dominated by leptons: $P_{\rm e,j} \gtrsim P_{\rm p,j}$. On the other hand, previous studies of small-scale jets in blazars and radio cores suggest a pair content (number of electrons and positrons per proton) of the order of $n_{\rm e}/n_{\rm p} \sim 20$. Assuming further that the particle composition of jets does not evolve beyond the blazar scales, we show that this implies an average random Lorentz factor of leptons in large-scale jets of $\barγ_{\rm e,j} \gtrsim 70(1+χ_{\rm p})(20n_{\rm p}/n_{\rm e})$, and that the protons should be mildly relativistic with $χ_{\rm p} \equiv (ε_{\rm p} + p_{\rm p})/ρ_{\rm p} c^2 \lesssim 2$, $p_{\rm p}$ the pressure of protons, $ε_{\rm p}$ the internal energy density of protons, and $ρ_{\rm p} c^2$ the rest-mass energy density of protons. We derive the necessary conditions for loading the inner jets by electron-positron pairs and proton-electron plasma, and provide arguments that heating of leptons in jets is dominated by magnetic reconnection.

astro-ph.HE↗

arXiv2020

Gradient Boosting Application in Forecasting of Performance Indicators Values for Measuring the Efficiency of Promotions in FMCG Retail

In the paper, a problem of forecasting promotion efficiency is raised. The authors propose a new approach, using the gradient boosting method for this task. Six performance indicators are introduced to capture the promotion effect. For each of them, within predefined groups of products, a model was trained. A description of using these models for forecasting and optimising promotion efficiency is provided. Data preparation and hyperparameters tuning processes are also described. The experiments were performed for three groups of products from a large grocery company.

cs.CY↗

arXiv2019

Comparison of SEDs of very massive radio-loud and radio-quiet AGN

The main objective of this work is to establish and interpret the dominant spectral components and their differences in radio-loud (RL) and radio-quiet (RQ) AGN with very massive black holes, and accreting at moderate rates. Such a sample is selected from the \textit{Swift}/BAT catalogue of AGN having determined optical spectra types and hosting black holes with masses $\rm > 10^{8.5}\,M_{\odot}$. We confirm our previous results, that radio loudness distribution of \textit{Swift}/BAT AGN is bimodal and that radio galaxies are about two times X-ray louder than their radio-quiet counterparts. We show that the average X-ray loudness (defined as a ratio of luminosity in the $14-195$ keV band to that at 12 $μ$m) of Type 1 and Type 2 AGN is very similar. This similarity holds for both RL and RQ subsamples and indicates negligible dependence of the observed X-ray luminosities on the inclination angle in both populations. In both the radiative output is dominated by mid-IR and hard X-ray components, and relatively weak UV luminosities indicate large amounts of dust in polar regions.

astro-ph.HE↗