Searcharxiv⌕ Search

arXiv subjects

Anjan Dutta

Publications and source records attributed to Anjan Dutta.

52 records · Page 3Linked to original sources
arXiv2020

A broadband look of the Accreting Millisecond X-ray Pulsar SAX J1748.9-2021 using AstroSat and XMM-Newton

SAX J1748.9-2021 is a transient accretion powered millisecond X-ray pulsar located in the Globular cluster NGC 6440. We report on the spectral and timing analysis of SAX J1748.9-2021 performed on AstroSat data taken during its faint and short outburst of 2017. We derived the best-fitting orbital solution for the 2017 outburst and obtained an average local spin frequency of 442.361098(3) Hz. The pulse profile obtained from 3-7 keV and 7-20 keV energy bands suggest constant fractional amplitude ~0.5% for fundamental component, contrary to previously observed energy pulse profile dependence. Our AstroSat observations revealed the source to be in a hard spectral state. The 1-50 keV spectrum from SXT and LAXPC on-board AstroSat can be well described with a single temperature blackbody and thermal Comptonization. Moreover, we found that the combined spectra from XMM-Newton (EPIC-PN) and AstroSat (SXT+LAXPC) indicated the presence of reflection features in the form of iron (Fe K$α$) line that we modeled with the reflection model xillvercp. One of the two X-ray burst observed during the AstroSat/LAXPC observation showed hard X-ray emission (>30 keV) due to Compton up-scattering of thermal photons by the hot corona. Time resolved analysis performed on the bursts revealed complex evolution in emission radius of blackbody for second burst suggestive of mild photospheric radius expansion.

astro-ph.HE↗
arXiv2020

Hierarchical stochastic graphlet embedding for graph-based pattern recognition

Despite being very successful within the pattern recognition and machine learning community, graph-based methods are often unusable because of the lack of mathematical operations defined in graph domain. Graph embedding, which maps graphs to a vectorial space, has been proposed as a way to tackle these difficulties enabling the use of standard machine learning techniques. However, it is well known that graph embedding functions usually suffer from the loss of structural information. In this paper, we consider the hierarchical structure of a graph as a way to mitigate this loss of information. The hierarchical structure is constructed by topologically clustering the graph nodes, and considering each cluster as a node in the upper hierarchical level. Once this hierarchical structure is constructed, we consider several configurations to define the mapping into a vector space given a classical graph embedding, in particular, we propose to make use of the Stochastic Graphlet Embedding (SGE). Broadly speaking, SGE produces a distribution of uniformly sampled low to high order graphlets as a way to embed graphs into the vector space. In what follows, the coarse-to-fine structure of a graph hierarchy and the statistics fetched by the SGE complements each other and includes important structural information with varied contexts. Altogether, these two techniques substantially cope with the usual information loss involved in graph embedding techniques, obtaining a more robust graph representation. This fact has been corroborated through a detailed experimental evaluation on various benchmark graph datasets, where we outperform the state-of-the-art methods.

cs.CV↗
arXiv2019

Semantically Tied Paired Cycle Consistency for Zero-Shot Sketch-based Image Retrieval

Zero-shot sketch-based image retrieval (SBIR) is an emerging task in computer vision, allowing to retrieve natural images relevant to sketch queries that might not been seen in the training phase. Existing works either require aligned sketch-image pairs or inefficient memory fusion layer for mapping the visual information to a semantic space. In this work, we propose a semantically aligned paired cycle-consistent generative (SEM-PCYC) model for zero-shot SBIR, where each branch maps the visual information to a common semantic space via an adversarial training. Each of these branches maintains a cycle consistency that only requires supervision at category levels, and avoids the need of highly-priced aligned sketch-image pairs. A classification criteria on the generators' outputs ensures the visual to semantic space mapping to be discriminating. Furthermore, we propose to combine textual and hierarchical side information via a feature selection auto-encoder that selects discriminating side information within a same end-to-end model. Our results demonstrate a significant boost in zero-shot SBIR performance over the state-of-the-art on the challenging Sketchy and TU-Berlin datasets.

cs.CV↗
arXiv2018

Stochastic Graphlet Embedding

Graph-based methods are known to be successful in many machine learning and pattern classification tasks. These methods consider semi-structured data as graphs where nodes correspond to primitives (parts, interest points, segments, etc.) and edges characterize the relationships between these primitives. However, these non-vectorial graph data cannot be straightforwardly plugged into off-the-shelf machine learning algorithms without a preliminary step of -- explicit/implicit -- graph vectorization and embedding. This embedding process should be resilient to intra-class graph variations while being highly discriminant. In this paper, we propose a novel high-order stochastic graphlet embedding (SGE) that maps graphs into vector spaces. Our main contribution includes a new stochastic search procedure that efficiently parses a given graph and extracts/samples unlimitedly high-order graphlets. We consider these graphlets, with increasing orders, to model local primitives as well as their increasingly complex interactions. In order to build our graph representation, we measure the distribution of these graphlets into a given graph, using particular hash functions that efficiently assign sampled graphlets into isomorphic sets with a very low probability of collision. When combined with maximum margin classifiers, these graphlet-based representations have positive impact on the performance of pattern comparison and recognition as corroborated through extensive experiments using standard benchmark databases.

cs.CV↗
arXiv2018

Study of the reflection spectra of SAX J1748.9-2021

We report the spectral analysis of accretion powered millisecond X-ray pulsar SAX J1748.9-2021 from the NuSTAR observation made during its 2015 outburst. The spectra displayed a broad emission line at $\sim 6.5$ keV with line width of $\sim 0.5$ keV and excess above $\sim 20$ keV due to Compton hump. The persistent emission of SAX J1748.9-2021 is described by a combination of soft thermal component with $kT =0.62^{+0.03}_{-0.05}$ keV and thermally Comptonized component with $kT_e=2.50^{+0.06}_{-0.03}$ keV reflected from the disc with reflection fraction of $0.30^{+0.08}_{-0.04}$. We have modeled the reflection spectrum with self-consistent model relxillCP and have found the inclination of the accretion disc to be $32.3^{^\circ +4.8}_{-4.7}$ and log $ξ$ is equal to $3.05^{+0.40}_{-0.34}$. We have obtained an upper limit of 57 km for inner disc radius; and the estimated magnetic field strength at the poles is less than $3.4 \times 10^9$ G. This kind of detailed investigation of reflection spectrum of SAX J1748.9-2021, especially above 10 keV, has been achieved only because of high statistics NuSTAR data of the source.

astro-ph.HE↗
arXiv2018

Spectral properties of MXB 1658-298 in the low/hard and high/soft state

We report results from a broadband spectral analysis of the low mass X-ray binary MXB 1658-298 with the Swift/XRT and NuSTAR observations made during its 2015-2017 outburst. The source showed different spectral states and accretion rates during this outburst. The source was in low/hard state during 2015; and it was in high/soft state during the 2016 NuSTAR observations. This is the first time that a comparison of the soft and hard spectral states during an outburst is being reported in MXB 1658-298. Compared with the observation of 2015, the X-ray luminosity was about four times higher during 2016. The hard state spectrum can be well described with models consisting of a single-temperature blackbody component along with Comptonized disc emission, or three component model comprising of multi-colour disc, single-temperature blackbody and thermal Comptonization components. The soft state spectrum can be described with blackbody or disc blackbody and Comptonization component, where the neutron star surface (or boundary layer) is the dominant source for the Comptonization seed photons. We have also found a link between the spectral state of the source and the Fe K absorber. The absorption features due to highly ionized Fe were observed only in the soft state. This suggests a probable connection between accretion disc wind (or atmosphere) and spectral state (or accretion state) of MXB 1658-298.

astro-ph.HE↗
arXiv2018

Learning Cross-Modal Deep Embeddings for Multi-Object Image Retrieval using Text and Sketch

In this work we introduce a cross modal image retrieval system that allows both text and sketch as input modalities for the query. A cross-modal deep network architecture is formulated to jointly model the sketch and text input modalities as well as the the image output modality, learning a common embedding between text and images and between sketches and images. In addition, an attention model is used to selectively focus the attention on the different objects of the image, allowing for retrieval with multiple objects in the query. Experiments show that the proposed method performs the best in both single and multiple object image retrieval in standard datasets.

cs.CV↗
arXiv2018

Graph Kernels based on High Order Graphlet Parsing and Hashing

Graph-based methods are known to be successful in many machine learning and pattern classification tasks. These methods consider semi-structured data as graphs where nodes correspond to primitives (parts, interest points, segments, etc.) and edges characterize the relationships between these primitives. However, these non-vectorial graph data cannot be straightforwardly plugged into off-the-shelf machine learning algorithms without a preliminary step of -- explicit/implicit -- graph vectorization and embedding. This embedding process should be resilient to intra-class graph variations while being highly discriminant. In this paper, we propose a novel high-order stochastic graphlet embedding (SGE) that maps graphs into vector spaces. Our main contribution includes a new stochastic search procedure that efficiently parses a given graph and extracts/samples unlimitedly high-order graphlets. We consider these graphlets, with increasing orders, to model local primitives as well as their increasingly complex interactions. In order to build our graph representation, we measure the distribution of these graphlets into a given graph, using particular hash functions that efficiently assign sampled graphlets into isomorphic sets with a very low probability of collision. When combined with maximum margin classifiers, these graphlet-based representations have positive impact on the performance of pattern comparison and recognition as corroborated through extensive experiments using standard benchmark databases.

cs.CV↗
arXiv2017

SigNet: Convolutional Siamese Network for Writer Independent Offline Signature Verification

Offline signature verification is one of the most challenging tasks in biometrics and document forensics. Unlike other verification problems, it needs to model minute but critical details between genuine and forged signatures, because a skilled falsification might often resembles the real signature with small deformation. This verification task is even harder in writer independent scenarios which is undeniably fiscal for realistic cases. In this paper, we model an offline writer independent signature verification task with a convolutional Siamese network. Siamese networks are twin networks with shared weights, which can be trained to learn a feature space where similar observations are placed in proximity. This is achieved by exposing the network to a pair of similar and dissimilar observations and minimizing the Euclidean distance between similar pairs while simultaneously maximizing it between dissimilar pairs. Experiments conducted on cross-domain datasets emphasize the capability of our network to model forgery in different languages (scripts) and handwriting styles. Moreover, our designed Siamese network, named SigNet, exceeds the state-of-the-art results on most of the benchmark signature datasets, which paves the way for further research in this direction.

cs.CV↗
arXiv2017

Indication of a massive circumbinary planet orbiting the Low Mass X-ray Binary MXB 1658-298

We present an X-ray timing analysis of the transient X-ray binary MXB 1658-298, using data obtained from the RXTE and XMM-Newton observatories. We have made 27 new mid eclipse time measurements from observations made during the two outbursts of the source. These new measurements have been combined with the previously known values to study long term changes in orbital period of the binary system. We have found that the mid-eclipse timing record of MXB 1658-298 is quite unusual. The long term evolution of mid-eclipse times indicates an overall orbital period decay with a time scale of -- 6.5(7) x 10^7 year. Over and above this orbital period decay, the O-C residual curve also shows a periodic residual on shorter timescales. This sinusoidal variation has an amplitude of ~9 lt-sec and a period of ~760 d. This is indicative of presence of a third body around the compact X-ray binary. The mass and orbital radius of the third body are estimated to lie in the range, 20.5-26.9 Jupiter mass and 750-860 lt-sec, respectively. If true, then it will be the most massive circumbinary planet and also the smallest period binary known to host a planet.

astro-ph.HE↗
arXiv2017

Product Graph-based Higher Order Contextual Similarities for Inexact Subgraph Matching

Many algorithms formulate graph matching as an optimization of an objective function of pairwise quantification of nodes and edges of two graphs to be matched. Pairwise measurements usually consider local attributes but disregard contextual information involved in graph structures. We address this issue by proposing contextual similarities between pairs of nodes. This is done by considering the tensor product graph (TPG) of two graphs to be matched, where each node is an ordered pair of nodes of the operand graphs. Contextual similarities between a pair of nodes are computed by accumulating weighted walks (normalized pairwise similarities) terminating at the corresponding paired node in TPG. Once the contextual similarities are obtained, we formulate subgraph matching as a node and edge selection problem in TPG. We use contextual similarities to construct an objective function and optimize it with a linear programming approach. Since random walk formulation through TPG takes into account higher order information, it is not a surprise that we obtain more reliable similarities and better discrimination among the nodes and edges. Experimental results shown on synthetic as well as real benchmarks illustrate that higher order contextual similarities add discriminating power and allow one to find approximate solutions to the subgraph matching problem.

cs.CV↗
arXiv2012

A New Scale for Attribute Dependency in Large Database Systems

Large, data centric applications are characterized by its different attributes. In modern day, a huge majority of the large data centric applications are based on relational model. The databases are collection of tables and every table consists of numbers of attributes. The data is accessed typically through SQL queries. The queries that are being executed could be analyzed for different types of optimizations. Analysis based on different attributes used in a set of query would guide the database administrators to enhance the speed of query execution. A better model in this context would help in predicting the nature of upcoming query set. An effective prediction model would guide in different applications of database, data warehouse, data mining etc. In this paper, a numeric scale has been proposed to enumerate the strength of associations between independent data attributes. The proposed scale is built based on some probabilistic analysis of the usage of the attributes in different queries. Thus this methodology aims to predict future usage of attributes based on the current usage.

cs.IR↗
arXiv2010

New measurement of orbital and spin period evolution of the Accretion Disk Corona source 4U 1822-37

4U 1822-37 is a Low Mass X-ray Binary (LMXB) system with an Accretion Disk Corona. We have obtained 16 new mid-eclipse time measurements of this source during the last 13 years using X- ray observations made with the RXTE-PCA, RXTE-ASM, Swift-XRT, XMM-Newton and Chandra observatories. These, along with the earlier known mid-eclipse times have been used to accurately determine the timescale for a change in the orbital period of 4U 1822-37. We have derived an orbital period Porb = 0.23210887(15) d, which is changing at the rate of \cdot Porb = 1.3(3) x 10-10 d d-1 (at T0 = MJD 45614). The timescale for a change in the orbital period is Porb/ \cdot Porb of 4.9(1.1) x 106 yr. We also report the detection of 0.59290132(11) s (at T0 = MJD 51975) X-ray pulsations from the source with a long term average \cdot Pspin of -2.481(4) x 10-12 s s-1, i.e., a spin-up time scale (Pspin/ \cdot Pspin) of 7578(13) yr. In view of these results, we have discussed various mechanisms that could be responsible for the orbital evolution in this system. Assuming the extreme case of conservative mass transfer, we have found that the measured \cdot Porb requires a large mass transfer rate of (4.2 - 5.2) x 10-8 M\odot yr-1 which together with the spin up rate implies a magnetic field strength in the range of (1-3) x 108 G. Using the long term RXTE-ASM light curve, we have found that the X-ray intensity of the source has decreased over the last 13 years by ? 40% and there are long term fluctuations at time scales of about a year. In addition to the long term intensity variation, we have also observed significant variation in the intensity during the eclipse.

astro-ph.HE↗
arXiv2009

Spectral and temporal changes associated with flux enhancement in 4U 1626-67

4U 1626-67 is an accretion powered X-ray pulsar that shows remarkably stable X-ray luminosity above hours timescale and gradual intensity variation on a few years timescale. Here we report a significant increase in the X-ray intensity in the long term RXTE-ASM light curve of 4U 1626-67. Similar enhancement in the X-ray flux has also been detected in the Swift-BAT light curve. The increase in the X-ray flux took place over a long period of about 100 days and there appears to be two episodes of flux enhancement. We have investigated the spectral and timing features of 4U 1626-67 during its current state of enhanced flux emission with data obtained from the Proportional Counter Array and the High-Energy X-ray Timing Explorer on board the Rossi X-ray Timing Explorer. The source has entered a new spin-up phase with a spin-up rate of 4.02(5) $\times$ 10$^{-13}$ Hz s$^{-1}$. The present spin-up rate is almost half of the earlier spin-up and spin-down trends. A significant excess in soft X-ray photon emission is observed during the enhanced flux state, which is similar to the energy spectrum obtained during the spin-up era of the pulsar before 1990. We also report detection of a significant broadening in the wings of the 130 mHz peak and a change in the shape of the continuum of the power spectrum.

astro-ph.HE↗
arXiv2009

Orbital X-ray modulation study of three Supergiant HMXBs

We present the orbital X-ray modulation study of three high mass X- ray binary systems, IGR J18027-2016, IGR J18483-0311 and IGR J16318-4848 using data obtained with RXTE-ASM, Swift-BAT and INTEGRAL-ISGRI. Using the long term light curves of the eclipsing HMXB IGR J18027-2016, obtained with Swift-BAT in the energy range 15-50 keV and INTEGRAL-ISGRI in the energy range 22-40 keV, we have determined three new mid eclipse times. The newly determined mid eclipse times together with the known values were used to derive an accurate value of the orbital period of 4.5693(4) d at MJD 52168 and an upper limit of 3.9(1.2)x10^-7 d d^-1 on the period derivative. We have also accurately determined an orbital period of 18.5482(88) d for the intermediate system IGR J18483-0311, which displays an unusual behaviour and shares many properties with the known SFXTs and persistent supergiant systems. This is a transient source and the outbursts occur intermittently at intervals of 18.55 d. Similarly, in the third supergiant system, IGR J16318-4848, we have found that the outbursts are separated by intervals of 80 d or its multiples, suggesting a possible orbital period.

astro-ph.HE↗
arXiv2009

Discovery of a short orbital period in the Supergiant Fast X-ray Transient IGR J16479-4514

We report here discovery of a 3.32 day orbital period in the Supergiant Fast X-ray Transient (SFXT) source IGR J16479-4514. Using the long term light curve of this source obtained with Swift-BAT in the energy range of 15-50 keV, we have clearly detected an orbital modulation including a full eclipse of duration ~0.6 day. In the hard X-ray band of the BAT instrument, the eclipse ingress and egress are rapid. We have also used the long term light curve obtained with the RXTE -ASM in the energy range of 1.5-12 keV. Taken independently, the detection of orbital modulation in the RXTE -ASM light curve is not significant. However, considering a clear detection of orbital modulation in the BAT light curve, we have used the ASM light curve for a more precise determination of the orbital period. IGR J16479-4514 has the shortest orbital period among the three SFXTs with measured/known orbital period. We discuss the implication of a short orbital period with the various mechanisms proposed to explain the transient nature of this class of sources.

astro-ph.HE↗