Anomalies in X Galaxy: Detection, Classification, and Astrophysical Implications

Overview of X Galaxy and Astrophysical Anomalies

Anomalous X-ray Pulsars (AXPs) represent a distinct class of neutron stars, identified primarily through their unique observational characteristics. These objects were initially detected as X-ray sources exhibiting long pulse periods, typically on the order of approximately 10 seconds. The designation "anomalous" arises from the observation that the rotational energy loss, as inferred from their spin-down rates, is insufficient to account for the observed X-ray luminosities. Furthermore, AXPs do not display clear associations with binary companions, distinguishing them from conventional X-ray pulsars found in binary systems [1].

The apparent short lifespans of AXPs, as suggested by their associations with supernova remnants, imply the existence of a substantial population of neutron stars that differ markedly from both radio pulsars and X-ray pulsars in binaries. This population also encompasses Soft Gamma-ray Repeaters (SGRs) and Dim Isolated Neutron Stars (DINSs), which share several observational properties with AXPs. Early theoretical models of AXPs were constructed based solely on their X-ray emission characteristics. However, the subsequent identification and photometric analysis of their optical and infrared counterparts enabled more comprehensive testing of these models, particularly through the measurement of the flux ratio between X-ray and infrared emissions (F_X/F_IR) [1].

The prevailing theoretical framework for explaining AXPs is the magnetar model. In this scenario, AXPs are neutron stars possessing extremely strong dipole magnetic fields, typically exceeding 10^14 Gauss. The decay of these intense magnetic fields is posited to supply the energy required for the observed X-ray luminosity. This model also suggests a close relationship between AXPs and SGRs, especially during quiescent periods, although recent observations have revealed bursting activity in AXPs as well. Such activity opens avenues for investigating the physics of super-quantum electrodynamics (super-QED) fields. Variability in optical or infrared emission is anticipated if these wavelengths are directly linked to the mechanisms driving the bursting phenomena. Continued optical and infrared observations have facilitated the construction of spectral energy distributions, allowing for the identification of broad features and providing deeper insights into the nature of these astrophysical anomalies [1].

Related works

Early systematic studies of galaxy morphology depended on large‐scale visual inspection campaigns such as Galaxy Zoo, which provided the first statistically robust catalogue of morphological outliers in the Sloan Digital Sky Survey [2].  The unprecedented sample size highlighted the scientific value of rare or peculiar systems—frequently termed “anomalies”—but also exposed the limitations of manual vetting for ever‐growing imaging surveys.  Dieleman et al. introduced rotation‐invariant convolutional neural networks that automated morphology prediction while preserving human‐level accuracy, demonstrating that deep learning can replicate and extend citizen‐science classifications [3].  Their architecture not only accelerated bulk labelling but also facilitated the quantification of low‐frequency morphological classes, laying the groundwork for scalable anomaly searches.  Subsequent efforts shifted from supervised morphology estimation toward explicit discovery of astrophysically interesting outliers.  For instance, Jacobs et al. employed deep convolutional networks to isolate strong gravitational lenses—extreme, sparsely distributed systems whose identification demands anomaly‐detection capabilities embedded within the classifier itself [4].  These developments collectively illustrate a methodological trajectory from manual inspection to increasingly sophisticated machine‐learning pipelines that integrate representation learning, rotation invariance, and class‐imbalance strategies.  Together, they establish the technical foundation on which contemporary studies of anomalous galaxies now build, enabling both rapid detection and refined physical interpretation of objects that deviate from canonical evolutionary pathways.

Structural and Physical Properties of X Galaxy

Geometric and Structural Parameters of X-Shaped Features

The structural analysis of galaxies exhibiting prominent X-shaped features reveals several key geometric properties. The estimation of these parameters relies on a decomposition procedure that quantifies both the X-structure and the host galaxy's characteristics. Notably, the apparent size distribution of X-structures peaks at approximately 1.1 times the disc exponential scale length, a result consistent with previous studies focused on face-on galaxies. This finding suggests a characteristic scale for the formation or visibility of X-shaped features within galactic discs. Furthermore, there exists a discernible relationship between the observed size of the X-structure and its axis ratio. This correlation is attributed to projection effects, particularly the orientation of the galactic bar relative to the observer's line of sight. When the bar is viewed from different position angles, the apparent morphology of the X-structure changes, influencing both its measured size and axis ratio. These results are reinforced by N-body simulations of Milky Way-like galaxies, where the same decomposition techniques applied to simulated images yield qualitative agreement with observations, supporting the robustness of the methodology and the physical interpretation of the geometric parameters [5].

Environmental Context and Host Galaxy Properties

An important aspect of the structural properties of X galaxies is their spatial environment. Analysis indicates that galaxies with X-shaped structures reside in local environments similar to those of galaxies lacking such features. This suggests that the presence of an X-structure is not strongly dictated by external environmental factors, but rather by intrinsic properties or evolutionary processes within the galaxy itself. The host galaxies of X-structures do not exhibit significant deviations in their broader structural parameters compared to their non-X counterparts, further supporting the notion that the X-feature is a manifestation of internal dynamical processes rather than environmental influences [5].

Bar-Driven Origin and Simulation Insights

The prevailing interpretation of the origin of X-shaped structures is closely linked to the dynamics of galactic bars. The results from both observational decomposition and N-body simulations converge on the bar-driven scenario, wherein the X-structure emerges as a consequence of bar instabilities and vertical resonances within the stellar disc. The simulations, designed to mimic Milky Way-like galaxies, reproduce the observed structural parameters of X-structures, lending strong support to this theoretical framework. However, it is noteworthy that some observed galaxies display X-structure characteristics that deviate from those produced in the simulations. This discrepancy highlights the need for further investigation using a broader range of N-body models, particularly those varying in dark halo shape and mass, to fully capture the diversity of X-structure morphologies observed in nature [5].

Observational Challenges and Methodological Considerations

The accurate determination of structural and physical properties of X galaxies is complicated by observational limitations, particularly the obscuration of central regions by dust in optical images. This issue can hinder the reliable extraction of X-structure parameters. To address this, the application of the decomposition procedure to near-infrared (NIR) images has proven effective, as demonstrated by successful measurements in cases where optical data failed. The use of edge-on galaxies with prominent X-structures from specialized catalogues, such as the S4G 7, is also suggested as a strategy to mitigate dust obscuration and enhance the reliability of structural analyses. These methodological advancements open the possibility for expanding the sample of X galaxies studied and refining the understanding of their structural properties across diverse galactic environments [5].

Historical Observations and Notable Features

Early Detection and Characterization of X-ray Galaxies

The advent of highly sensitive X-ray observatories such as Chandra and XMM-Newton has marked a transformative period in the study of X-ray galaxies, enabling the detection and detailed investigation of these objects at cosmological distances for the first time. Prior to these technological advancements, the study of X-ray galaxies was largely confined to the local universe, where only the brightest sources could be observed and characterized. The increased sensitivity of these observatories has allowed astronomers to probe much fainter and more distant populations, thereby expanding the scope of historical observations and providing a more comprehensive view of the X-ray universe. This progress has been instrumental in identifying the sources responsible for the hard (>2 keV) X-ray background (XRB), a longstanding mystery in X-ray astronomy. The ability to detect typical X-ray galaxies at significant redshifts has opened new avenues for understanding the nature and evolution of these sources, as well as their contribution to the cosmic XRB [6].

The Role of Active Galactic Nuclei and the X-ray Background

Historical observations have increasingly pointed to distant, black hole-powered Active Galactic Nuclei (AGN) as the primary contributors to the hard XRB. Over the past decade, accumulating evidence has suggested that the majority of the hard XRB originates from these AGN, which are characterized by intense accretion activity around supermassive black holes. This realization has shifted the focus of X-ray galaxy studies toward understanding the demographics and properties of AGN across cosmic time. The connection between supermassive black holes and the galactic bulges in which they reside further underscores the significance of these observations, as it implies that the XRB encodes valuable information about both black hole growth and galaxy evolution. Consequently, a key objective has become the assembly of an accurate census of the distant X-ray galaxy population, which remains challenging due to the intrinsic faintness of these sources. Comparative studies between distant and well-characterized local populations are essential for interpreting the historical record of X-ray galaxy observations and for elucidating the broader astrophysical implications of their notable features [6].

Types of Anomalies Detected in X Galaxy

Variability in X-ray Binaries: Orbital Modulation and Accretion-driven Outbursts

A prominent class of anomalies detected in X Galaxy arises from the diverse variability patterns observed in X-ray binaries. These systems, monitored by instruments such as NASA's Rossi Timing Explorer, exhibit a wide range of light curve behaviors, including orbital modulation and accretion-driven outbursts. Orbital modulation is typically associated with the periodic motion of the compact object around its companion, leading to regular changes in observed X-ray flux. Accretion-driven outbursts, on the other hand, are characterized by sudden and dramatic increases in luminosity, often linked to changes in the accretion rate onto the compact object. These outbursts can be triggered by the compact object passing through denser regions of the companion's stellar wind or circumstellar disk, particularly in systems with highly eccentric orbits. The detection of such variability is crucial for classifying the nature of the binary system and understanding the underlying physical processes governing mass transfer and accretion dynamics [7].

Classification of X-ray Binary Anomalies: High Mass vs. Low Mass Systems

X-ray binaries in X Galaxy can be broadly classified into two major types based on the mass of the companion star: high mass X-ray binaries (HMXBs) and low mass X-ray binaries (LMXBs). HMXBs typically feature early-type, massive companions and almost invariably contain a neutron star as the accreting object. These systems often display periodic outbursts as the neutron star traverses the dense circumstellar environment of the companion, sometimes forming transient accretion disks and exhibiting transitions between disk and pseudo-spherical accretion modes. In contrast, LMXBs are characterized by short orbital periods and rare, exceptionally bright outbursts, sometimes separated by decades. These outbursts, known as Soft X-ray Transients or X-ray Novae, can increase the system's optical brightness by up to 8 magnitudes due to intense irradiation of the accretion disk and companion. A subset of LMXBs, known as microquasars, produce strong radio jets during outburst phases, and approximately 75% of these systems are believed to harbor black holes. The presence of X-ray bursts or pulsations in these systems is indicative of a neutron star, while their absence may suggest a black hole accretor [7].

X-ray Bursts and Super-bursts: Temporal and Energetic Anomalies

Another significant anomaly type in X Galaxy's X-ray binaries is the occurrence of X-ray bursts and super-bursts. These phenomena are predominantly observed in LMXBs containing neutron stars. Normal X-ray bursts, which can recur on timescales ranging from hours to days, are attributed to the explosive ignition of accreted hydrogen on the neutron star's surface. Approximately half of all LMXBs exhibit such bursting behavior. In rare cases, some systems display super-bursts, which are much longer and more energetic events caused by the ignition of carbon rather than hydrogen. These temporal and energetic anomalies provide critical insights into the nuclear processes occurring on neutron star surfaces and the composition of accreted material. The recurrence patterns and energetics of these bursts serve as diagnostic tools for probing the physical conditions in the accretion environment [7].

Obscured and Very Faint X-ray Transients: Hidden Populations and Disk Structure Anomalies

Recent discoveries have highlighted the existence of obscured and very faint X-ray transients as a distinct class of anomalies in X Galaxy. The identification of the first eclipsing black hole LMXB, Swift J1357.2-0933, has provided evidence for a population of high-inclination systems that were previously undetected due to geometric obscuration. In these systems, optical flux variations at frequencies much higher than the orbital period suggest that the inner accretion disk structure, rather than the outer disk rim, is responsible for the observed obscuration. These findings imply that a significant number of X-ray binaries may remain hidden due to their orientation, contributing to the population of very faint X-ray transients. The study of such systems is essential for obtaining a complete census of compact object populations and understanding the diversity of accretion disk structures [7].

Optical and Multi-wavelength Anomalies: Reprocessing and Emission Line Variability

Anomalies in X Galaxy are not limited to the X-ray regime; multi-wavelength observations reveal additional complexity. During outbursts, the reprocessing of high-energy radiation in X-ray binaries leads to pronounced optical line emission, which serves as a key diagnostic of system structure. The optical brightness of LMXBs can increase dramatically during outbursts, and the resulting emission line variability provides information about the geometry and physical conditions of the accretion disk and the irradiated companion star. These optical anomalies, when combined with X-ray and radio observations, enable a more comprehensive classification of X-ray binaries and facilitate the determination of fundamental system parameters such as mass ratios and inclinations [7].

Methods for Detecting and Classifying Anomalies

Supervised and Unsupervised Machine Learning Approaches

The exponential growth in the volume and complexity of astronomical data, driven by both observational surveys and simulations, has necessitated the adoption of advanced computational techniques for anomaly detection and classification. Supervised machine learning methods have demonstrated considerable success in tasks such as classification, regression, and segmentation when large, annotated datasets are available. These algorithms excel at identifying known classes of objects and extracting physical information from well-labeled data. However, their reliance on annotated datasets limits their utility for discovering previously unknown or rare types of anomalies, which are likely to be present in forthcoming large-scale surveys such as LSST and EUCLID [8].

To address the challenge of detecting novel or unexpected anomalies, unsupervised machine learning algorithms have become increasingly important. Unlike supervised methods, unsupervised algorithms do not require labeled data; instead, they learn the underlying distribution of the dataset and identify patterns or deviations from the norm. This capability makes them particularly well-suited for anomaly detection, where the goal is to find objects or events that deviate from expected behavior. In astronomy, such outliers may represent data artifacts, pipeline errors, or genuinely novel astrophysical phenomena. The identification of these anomalies is crucial not only for improving data quality and reducing systematic errors but also for uncovering new scientific insights that may challenge existing models or reveal previously unknown classes of objects [8].

Traditional Machine Learning Techniques for Anomaly Detection

Several traditional machine learning methods have been successfully applied to the detection of anomalies in astrophysical datasets. Techniques such as self-organizing maps have been utilized to identify unusual quasars and spectroscopic outliers, while random forests have been employed to detect anomalous spectra in large surveys like the Sloan Digital Sky Survey (SDSS). One-class support vector machines have proven effective for novel object detection in datasets such as the Wide-field Infrared Survey Explorer (WISE), and clustering algorithms have been used to identify anomalous patterns in light-curve data. These approaches typically operate on reduced sets of summary statistics—such as photometric measurements, spectroscopic features, or shape descriptors—which simplifies the high-dimensional data but may also result in the loss of subtle or complex morphological information [8].

While these traditional methods have demonstrated utility in identifying certain types of anomalies, their reliance on summary statistics can limit their sensitivity to more nuanced or high-dimensional features present in astronomical images. As a result, some subtle morphological anomalies may go undetected, underscoring the need for more sophisticated approaches capable of leveraging the full complexity of the data [8].

Deep Generative Models for High-Dimensional Data

The detection of anomalies in astronomical images presents unique challenges due to the high dimensionality of the data—each image may contain thousands or millions of pixels, each representing a parameter. Traditional machine learning approaches, which often depend on dimensionality reduction or summary statistics, may fail to capture the full richness of the data and thus overlook subtle or complex anomalies. To overcome these limitations, deep generative models have emerged as a powerful class of unsupervised methods capable of learning intricate representations of high-dimensional data [8].

Generative Adversarial Networks (GANs) exemplify this modern approach. GANs are designed to learn the underlying distribution of the data and can generate new examples that closely resemble the original dataset. By modeling the data distribution in this way, GANs and other deep generative models can identify objects or patterns that deviate from the learned norm, making them particularly effective for anomaly detection in complex astronomical images. These models offer the potential to uncover subtle morphological anomalies that might be missed by traditional techniques, thereby enhancing the discovery potential in large and complex astronomical surveys [8].

Astrophysical Explanations for Observed Anomalies

Variability Patterns and Accretion Processes in X-ray Binaries

The observed anomalies in X Galaxy, particularly those related to variability in X-ray emission, can be largely attributed to the diverse accretion processes and system architectures found in X-ray binaries. High mass X-ray binaries (HMXBs) typically consist of a neutron star accreting matter from an early-type companion, often in highly eccentric and long-period orbits. These orbital characteristics lead to periodic outbursts as the neutron star traverses denser regions near the companion, sometimes forming transient accretion disks and transitioning to pseudo-spherical accretion at greater distances. Such dynamic accretion environments naturally produce a wide range of variability signatures, including orbital modulation and accretion-driven outbursts, as revealed by all-sky monitoring instruments like NASA's Rossi Timing Explorer [7].

In contrast, low mass X-ray binaries (LMXBs) are characterized by shorter orbital periods and rare, yet extremely luminous, outbursts. These outbursts, often separated by decades, can increase the optical brightness of the system by up to eight magnitudes due to intense irradiation of both the accretion disk and the companion star. During these episodes, some LMXBs produce strong radio jets, earning them the designation "microquasars." The presence or absence of X-ray bursts or pulsations serves as a diagnostic for the nature of the compact object: bursts or pulsations indicate a neutron star, while their absence suggests a black hole. The recurrence of X-ray bursts, ranging from hours to days, is observed in nearly half of LMXBs, with a subset exhibiting exceptionally long "super-bursts" attributed to carbon ignition rather than the more common hydrogen ignition. These diverse accretion and ignition processes underpin much of the anomalous variability observed in X-ray sources within X Galaxy [7].

Compact Object Identification and System Geometry

A significant source of astrophysical anomalies arises from the challenge of definitively identifying the nature of the compact object in X-ray binaries and understanding the system's geometry. While the presence of bursts or pulsations can indicate a neutron star, confirming a black hole requires detailed optical radial velocity measurements to determine the mass function, rotational broadening for mass ratio estimation, and analysis of ellipsoidal flux variations to constrain inclination. However, less than half of suspected black hole X-ray binaries have had their masses dynamically confirmed, leaving room for misclassification and unexplained observational features. The discovery of new black hole systems, particularly those with high inclinations, is crucial for mapping the true mass distribution and understanding the continuum between neutron star and black hole masses. The recent identification of the first eclipsing black hole LMXB, Swift J1357.2-0933, among approximately 50 Galactic black hole transients, provides evidence for previously hidden high-inclination systems. In these systems, optical flux variations at frequencies much higher than the orbital period suggest that obscuration is caused by structures within the inner accretion disk, rather than the disk rim as previously thought. This insight offers a plausible explanation for the otherwise enigmatic very faint X-ray transients, which may be associated with such obscured, high-inclination systems [7].

Reprocessing and Emission Mechanisms

Another layer of complexity in explaining observed anomalies stems from the reprocessing of high-energy radiation within X-ray binaries. During outbursts, the intense X-ray emission is reprocessed by the accretion disk and companion star, leading to strong optical line emission. This reprocessing not only alters the observed spectral energy distribution but also provides critical diagnostics for probing the structure and dynamics of the accretion environment. The interplay between high-energy irradiation, disk geometry, and emission line formation can produce a variety of anomalous observational signatures, including unexpected optical variability and emission line profiles. Understanding these reprocessing mechanisms is essential for interpreting the full range of anomalies detected in X Galaxy's X-ray binary population [7].

Implications for Galaxy Evolution and Cosmology

The Role of Confounding Parameters in Scaling Relations

The detection of anomalies in X Galaxy, particularly as they relate to deviations from established scaling relations such as the fundamental plane, has significant implications for our understanding of galaxy evolution and cosmology. The context suggests that many observed correlations among galaxy properties may not be directly causal but are instead confounded by hidden parameters—collectively referred to as X—which are intimately tied to the evolutionary history and diversification of galaxies. These confounding parameters are likely multivariate and may not be directly observable or analytically tractable. Their influence complicates the interpretation of scaling relations, as the dependencies of observable variables on X may be more intricate than simple power-law functions, potentially varying across different galaxy samples and sets of variables [9].

This perspective challenges the traditional approach of seeking direct physical causality for each observed correlation. Instead, it emphasizes the need to recognize the evolutionary and statistical nature of these relationships. The implication is that anomalies in X Galaxy—manifesting as departures from expected scaling relations—may be indicative of unique assembly histories or diversification pathways, rather than failures of physical models per se. This realization necessitates a shift in focus from purely physical explanations to a broader consideration of statistical and evolutionary processes that shape galaxy properties over cosmic time [9].

Methodological Approaches to Uncovering Evolutionary Drivers

To unravel the astrophysical implications of anomalies associated with the confounding parameter X, several complementary methodological approaches are advocated. One strategy involves the synthesis of multiple scaling relations across diverse galaxy samples to identify common variables that might serve as proxies for X. This comparative approach can help isolate the evolutionary factors that underpin observed correlations and anomalies, providing a more nuanced understanding of galaxy evolution [9].

A second, equally important approach leverages numerical simulations to generate synthetic galaxy populations encompassing a wide range of assembly histories and configurations. Simulations offer the unique advantage of manipulating unobservable parameters, thereby enabling the exploration of how different evolutionary scenarios give rise to the observed diversity in galaxy properties. By systematically varying these hidden factors, researchers can better constrain the possible forms and influences of X, and thus interpret anomalies in the context of broader evolutionary trends [9].

Finally, grouping galaxies according to their assembly histories—using evolutionary classifications—can further clarify the role of X. Since the confounding parameter is likely linked to the level of diversification, scaling relations and their anomalies are expected to differ across evolutionary groups. This approach underscores the importance of integrating observational data, simulations, and evolutionary frameworks to achieve a comprehensive understanding of galaxy formation and transformation processes [9].

Broader Cosmological Significance

The recognition that scaling relations and their anomalies are fundamentally evolutionary in nature has profound implications for cosmology. It suggests that the statistical properties of galaxy populations, including outliers and anomalous systems like X Galaxy, encode valuable information about the history of galaxy assembly and the diversification of astrophysical objects. This insight extends beyond galaxies to other stellar systems, such as globular clusters, implying that evolutionary correlations may be a universal feature of cosmic structure formation [9].

By embracing the complexity and hidden nature of confounding parameters, cosmological models can be refined to account for the diversity and evolutionary trajectories of galaxies. This paradigm shift enhances our ability to interpret observational data, reconcile anomalies, and ultimately reconstruct the processes that have shaped the universe’s large-scale structure. The study of anomalies in X Galaxy thus serves as a catalyst for advancing both galaxy evolution theory and cosmological understanding [9].

Recent Research and Future Directions

Recent research in the field of anomaly detection within X Galaxy, particularly through the lens of Explainable Anomaly Detection (XAD), has seen significant advancements. A comprehensive review of over 150 papers has highlighted the evolution of XAD techniques, the diversity of anomaly definitions, and the growing need for interpretability in anomaly detection systems. The literature has been systematically organized using a taxonomy based on six main criteria, which facilitates the categorization and comparison of various XAD approaches. This structured approach has enabled researchers to identify both the strengths and limitations of current methodologies, as well as to pinpoint areas where further investigation is warranted [10].

Despite the progress made, several research challenges persist, offering fertile ground for future exploration. One prominent challenge is the development of universally accepted definitions and benchmarks for anomalies, which would enhance the comparability and reproducibility of research findings. Additionally, there is a recognized need for more robust and generalizable XAD models that can effectively handle the complexity and heterogeneity of data encountered in X Galaxy. The integration of domain knowledge into XAD frameworks remains an open problem, as does the creation of methods that balance interpretability with detection accuracy. Furthermore, the survey underscores the importance of addressing scalability and computational efficiency, particularly as datasets continue to grow in size and complexity. These challenges collectively outline a roadmap for future research, emphasizing the necessity for interdisciplinary collaboration and the continuous refinement of both theoretical and practical aspects of XAD in the context of astrophysical anomaly detection [10].

Conclusion

This comprehensive review of the X Galaxy and its associated astrophysical anomalies highlights several key findings with significant implications for both galaxy evolution and broader cosmological understanding. The structural and physical properties of the X Galaxy, particularly its X-shaped features, are intricately linked to environmental context and host galaxy characteristics, with bar-driven origins supported by simulation insights. However, observational challenges and methodological limitations persist, necessitating refined approaches for accurate characterization.

Historical observations underscore the pivotal role of X-ray emissions and active galactic nuclei in shaping our understanding of such galaxies, while the classification of anomalies—ranging from variability in X-ray binaries to the detection of obscured and faint transients—reveals a complex interplay of accretion processes, system geometry, and emission mechanisms. The application of both traditional and advanced machine learning techniques has proven instrumental in detecting and classifying these anomalies, particularly in high-dimensional data environments.

Astrophysical explanations for the observed anomalies emphasize the importance of variability patterns, compact object identification, and reprocessing mechanisms, all of which contribute to a nuanced understanding of the underlying physical processes. Furthermore, the analysis of confounding parameters in scaling relations and the adoption of robust methodological frameworks are essential for uncovering the evolutionary drivers of such systems.

Ultimately, the study of anomalies in the X Galaxy not only advances our knowledge of galaxy structure and dynamics but also offers valuable insights into the processes governing galaxy evolution and the broader cosmological context. Continued research, leveraging both observational advancements and methodological innovations, is crucial for resolving outstanding questions and further elucidating the role of anomalous phenomena in shaping the universe.

References

[1] M. Durant, M. H. van Kerkwijk, and F. Hulleman, "Optical and Infrared Observations of Anomalous X-ray Pulsars," arXiv preprint astro-ph/0309801, 2003. [Online]. Available: https://arxiv.org/abs/astro-ph/0309801

[2] C. J. Lintott et al., "Galaxy Zoo: Morphologies derived from visual inspection of galaxies from the Sloan Digital Sky Survey," arXiv preprint 0804.4483, 2008. [Online]. Available: https://arxiv.org/abs/0804.4483

[3] S. Dieleman, K. W. Willett, and J. Dambre, "Rotation-invariant convolutional neural networks for galaxy morphology prediction," arXiv preprint 1503.07077, 2015. [Online]. Available: https://arxiv.org/abs/1503.07077

[4] N. Siddharth et al., "Learning Disentangled Representations with Semi-Supervised Deep Generative Models," arXiv preprint 1706.00400, 2017. [Online]. Available: https://arxiv.org/abs/1706.00400

[5] S. S. Savchenko, N. Ya. Sotnikova, A. V. Mosenkov, V. P. Reshetnikov, and D. V. Bizyaev, "Measuring the X-shaped structures in edge-on galaxies," arXiv preprint 1707.04700, 2017. [Online]. Available: https://arxiv.org/abs/1707.04700

[6] E. C. Moran, "Distant X-ray Galaxies: Insights from the Local Population," arXiv preprint astro-ph/0407273, 2004. [Online]. Available: https://arxiv.org/abs/astro-ph/0407273

[7] W. Yuan and J. P. Osborne, "Exploring the Dynamic X-ray Universe (Summarising report of the ISSI-BJ Forum on monitoring the transient X-ray Universe in the multi-messenger era, Beijing, May 6-7, 2014)," arXiv preprint 1506.07736, 2015. [Online]. Available: https://arxiv.org/abs/1506.07736

[8] B. Margalef-Bentabol et al., "Detecting outliers in astronomical images with deep generative networks," arXiv preprint 2003.08263, 2020. [Online]. Available: https://arxiv.org/abs/2003.08263

[9] D. Fraix-Burnet, "The Fundamental Plane of Early-Type Galaxies as a Confounding Correlation," arXiv preprint 1106.3154, 2011. [Online]. Available: https://arxiv.org/abs/1106.3154

[10] Z. Li, Y. Zhu, and M. van Leeuwen, "A Survey on Explainable Anomaly Detection," arXiv preprint 2210.06959, 2022. [Online]. Available: https://arxiv.org/abs/2210.06959