arXiv subjects

Amitava Datta

Publications and source records attributed to Amitava Datta.

At least 37 records · Page 2Linked to original sources

arXiv2021

BERT based classification system for detecting rumours on Twitter

The role of social media in opinion formation has far-reaching implications in all spheres of society. Though social media provide platforms for expressing news and views, it is hard to control the quality of posts due to the sheer volumes of posts on platforms like Twitter and Facebook. Misinformation and rumours have lasting effects on society, as they tend to influence people's opinions and also may motivate people to act irrationally. It is therefore very important to detect and remove rumours from these platforms. The only way to prevent the spread of rumours is through automatic detection and classification of social media posts. Our focus in this paper is the Twitter social medium, as it is relatively easy to collect data from Twitter. The majority of previous studies used supervised learning approaches to classify rumours on Twitter. These approaches rely on feature extraction to obtain both content and context features from the text of tweets to distinguish rumours and non-rumours. Manually extracting features however is time-consuming considering the volume of tweets. We propose a novel approach to deal with this problem by utilising sentence embedding using BERT to identify rumours on Twitter, rather than the usual feature extraction techniques. We use sentence embedding using BERT to represent each tweet's sentences into a vector according to the contextual meaning of the tweet. We classify those vectors into rumours or non-rumours by using various supervised learning techniques. Our BERT based models improved the accuracy by approximately 10% as compared to previous methods.

cs.LG↗

arXiv2021

Estimating galaxy masses from kinematics of globular cluster systems: a new method based on deep learning

We present a new method by which the total masses of galaxies including dark matter can be estimated from the kinematics of their globular cluster systems (GCSs). In the proposed method, we apply the convolutional neural networks (CNNs) to the two-dimensional (2D) maps of line-of-sight-velocities ($V$) and velocity dispersions ($σ$) of GCSs predicted from numerical simulations of disk and elliptical galaxies. In this method, we first train the CNN using either only a larger number ($\sim 200,000$) of the synthesized 2D maps of $σ$ ("one-channel") or those of both $σ$ and $V$ ("two-channel"). Then we use the CNN to predict the total masses of galaxies (i.e., test the CNN) for the totally unknown dataset that is not used in training the CNN. The principal results show that overall accuracy for one-channel and two-channel data is 97.6\% and 97.8\% respectively, which suggests that the new method is promising. The mean absolute errors (MAEs) for one-channel and two-channel data are 0.288 and 0.275 respectively, and the value of root mean square errors (RMSEs) are 0.539 and 0.51 for one-channel and two-channel respectively. These smaller MAEs and RMSEs for two-channel data (i.e., better performance) suggest that the new method can properly consider the global rotation of GCSs in the mass estimation. We also applied our proposed method to real data collected from observations of NGC 3115 to compare the total mass predicted by our proposed method and other popular methods from the literature.

astro-ph.CO↗

arXiv2021

Scheduling Applications on Containers Based on Dependency of The Applications

Cloud computing technology has been one of the most critical developments in provisioning both hardware and software infrastructure in recent years. Container technology is a new cloud technology that boosts the booting of applications, increases the ability to deploy applications on containers and improves the host machine resource sharing. Thus, enhancing a cloud container system needs a robust algorithm that deploys the applications efficiently. Most of the schedulers associated with container technology are focused on load balancing for increasing container performance. The traffic over networks plays a significant role in the performance of containers. Container deployment considering only load balancing may not be the best scheduling strategy due to the dependency between the applications that might be deployed in different pods (zones) in the container's cloud. This project aims to develop an algorithm that deploys applications into containers by considering the dependencies between applications as well as load balancing. The proposed algorithm performs better in terms of improving the throughput and reducing the network traffic as compared to state-of-the-art container scheduling algorithms.

cs.DC↗

arXiv2020

Automatic detection of abnormal EEG signals using wavelet feature extraction and gradient boosting decision tree

Electroencephalography is frequently used for diagnostic evaluation of various brain-related disorders due to its excellent resolution, non-invasive nature and low cost. However, manual analysis of EEG signals could be strenuous and a time-consuming process for experts. It requires long training time for physicians to develop expertise in it and additionally experts have low inter-rater agreement (IRA) among themselves. Therefore, many Computer Aided Diagnostic (CAD) based studies have considered the automation of interpreting EEG signals to alleviate the workload and support the final diagnosis. In this paper, we present an automatic binary classification framework for brain signals in multichannel EEG recordings. We propose to use Wavelet Packet Decomposition (WPD) techniques to decompose the EEG signals into frequency sub-bands and extract a set of statistical features from each of the selected coefficients. Moreover, we propose a novel method to reduce the dimension of the feature space without compromising the quality of the extracted features. The extracted features are classified using different Gradient Boosting Decision Tree (GBDT) based classification frameworks, which are CatBoost, XGBoost and LightGBM. We used Temple University Hospital EEG Abnormal Corpus V2.0.0 to test our proposed technique. We found that CatBoost classifier achieves the binary classification accuracy of 87.68%, and outperforms state-of-the-art techniques on the same dataset by more than 1% in accuracy and more than 3% in sensitivity. The obtained results in this research provide important insights into the usefulness of WPD feature extraction and GBDT classifiers for EEG classification.

eess.SP↗

arXiv2020

Hit ratio: An Evaluation Metric for Hashtag Recommendation

Hashtag recommendation is a crucial task, especially with an increase of interest in using social media platforms such as Twitter in the last decade. Hashtag recommendation systems automatically suggest hashtags to a user while writing a tweet. Most of the research in the area of hashtag recommendation have used classical metrics such as hit rate, precision, recall, and F1-score to measure the accuracy of hashtag recommendation systems. These metrics are based on the exact match of the recommended hashtags with their corresponding ground truth. However, it is not clear how adequate these metrics to evaluate hashtag recommendation. The research question that we are interested in seeking an answer is: are these metrics adequate for evaluating hashtag recommendation systems when the numbers of ground truth hashtags in tweets are highly variable? In this paper, we propose a new metric which we call hit ratio for hashtag recommendation. Extensive evaluation through hypothetical examples and real-world application across a range of hashtag recommendation models indicate that the hit ratio is a useful metric. A comparison of hit ratio with the classical evaluation metrics reveals their limitations.

cs.IR↗

arXiv2019

Using Deep Learning to Localize Gravitational Wave Sources

In this paper, we report on the construction of a deep Artificial Neural Network (ANN) to localize simulated gravitational wave signals in the sky with high accuracy. We have modelled the sky as a sphere and have considered cases where the sphere is divided into 18, 50, 128, 1024, 2048 and 4096 sectors. The sky direction of the gravitational wave source is estimated by classifying the signal into one of these sectors based on it's right ascension and declination values for each of these cases. In order to do this, we have injected simulated binary black hole gravitational wave signals of component masses sampled uniformly between 30-80 solar mass into Gaussian noise and used the whitened strain values to obtain the input features for training our ANN. We input features such as the delays in arrival times, phase differences and amplitude ratios at each of the three detectors Hanford, Livingston and Virgo, from the raw time-domain strain values as well as from analytical versions of these signals, obtained through Hilbert transformation. We show that our model is able to classify gravitational wave samples, not used in the training process, into their correct sectors with very high accuracy (>90%) for coarse angular resolution using 18, 50 and 128 sectors. We also test our localization on test samples with injection parameters of the published LIGO binary black hole merger events GW150914, GW170818 and GW170823 for 1024, 2048 and 4096 sectors and compare the result with that from BAYESTAR and Parameter Estimation (PE). In addition, we report that the time taken by our model to localize one GW signal is around 0.018 secs on 14 Intel Xeon CPU cores.

astro-ph.IM↗

arXiv2019

The past, present and future of the heavier electroweakinos in the light of LHC and other data

The aim of this paper is to showcase the novel multilepton ($nl + E\!\!\!\!/_T$, n = 3 - 5) signals, hitherto unexplored at the LHC, arising from the heavier electroweakinos, in several generic pMSSMs at the upcoming LHC experiments. We first briefly review our old constraints on the full electroweakino sector of these models, containing both lighter and heavier sparticles, using the ATLAS trilepton data from the LHC Run I. Next we derive new stronger constraints on this sector for the first time using the ATLAS Run II data. We identify some benchmark points and explore the prospect of observing multilepton events in future LHC experiments. Our focus is on the channels with $n > 3$ which are the hallmarks of the heavier electroweakinos. If the spectrum of the lighter electroweakinos is compressed, these signals might very well be the discovery channels of the electroweakinos at the high luminosity LHC. We also discuss the implications of the new LHC constraints for the observed dark matter relic density of the universe, the measured value of the anomalous magnetic moment of the muon and the dark matter direct detection experiments.

hep-ph↗

arXiv2018

Stable target opinion through power law bias in information exchange

We study a model of binary decision making when a certain population of agents is initially seeded with two different opinions, `$+$' and `$-$', with fractions $p_1$ and $p_2$ respectively, $p_1+p_2=1$. Individuals can reverse their initial opinion only once based on this information exchange. We study this model on a completely connected network, where any pair of agents can exchange information, and a two-dimensional square lattice with periodic boundary conditions, where information exchange is possible only between the nearest neighbors. We propose a model in which each agent maintains two counters of opposite opinions and accepts opinions of other agents with a power law bias until a threshold is reached, when they fix their final opinion. Our model is inspired by the study of negativity bias and positive-negative asymmetry known in the psychology literature for a long time. Our model can achieve stable intermediate mix of positive and negative opinions in a population. In particular, we show that it is possible to achieve close to any fraction $p_3, 0\leq p_3\leq 1$, of `$-$' opinion starting from an initial fraction $p_1$ of `$-$' opinion by applying a bias through adjusting the power law exponent of $p_3$.

physics.soc-ph↗

arXiv2018

Multilepton signals of heavier electroweakinos at the LHC

As sequel to a recent paper we examine the phenomenology of the full electroweakino sector of the pMSSM without invoking the adhoc but often employed assumption that the heavier ones are decoupled. We identify several generic models which illustrate the importance of the heavier electroweakinos and constrain them with the LHC $3l$ + ${E\!\!\!\!/_T}$ data. The constraints are usually stronger than that for decoupled heavier electroweakinos indicating that the LHC data is already sensitive to their presence. We also take into account the constraints from the observed dark matter relic density of the universe and precisely measured anomalous magnetic moment of the muon. Using the allowed parameter space thus obtained, we show that in addition to the conventional $3l$ + ${E\!\!\!\!/_T}$ signatures novel multilepton ($ml$) + ${E\!\!\!\!/_T}$ final states with $m > 3$, which are not viable in models with lighter electroweakinos only, can be observed before the next long shut down of the LHC.

hep-ph↗

arXiv2017

Double Transition in a Model of Oscillating Percolation

Two distinct transition points have been observed in a problem of lattice percolation studied using a system of pulsating discs. Sites on a regular lattice are occupied by circular discs whose radii vary sinusoidally within $[0,R_0]$ starting from a random distribution of phase angles. A lattice bond is said to be connected when its two end discs overlap with each other. Depending on the difference of the phase angles of these discs a bond may be termed as dead or live. While a dead bond can never be connected, a live bond is connected at least once in a complete time period. Two different time scales can be associated with such a system, leading to two transition points. Namely, a percolation transition occurs at $R_{0c} =0.908$ when a spanning cluster of connected bonds emerges in the system. Here, information propagates across the system instantly, i.e., with infinite speed. Secondly, there exists another transition point $R_0^* = 0.5907$ where the giant cluster of live bonds spans the lattice. In this case the information takes finite time to propagate across the system through the dynamical evolution of finite size clusters. This passage time diverges as $R_0 \to R_0^*$ from above. Both the transitions exhibit the critical behavior of ordinary percolation transition. The entire scenario is robust with respect to the distribution of frequencies of the individual discs. This study may be relevant in the context of wireless sensor networks.

cond-mat.stat-mech↗

arXiv2016

Exploring collider signatures of the inert Higgs doublet model

We revisit the multilepton ($ml$) + ${E\!\!\!\!/_T}$ + $X$ signatures of the Inert Doublet Model (IDM) of dark matter in future LHC experiments for $m =$ 3, 4 and simulate, for the first time, the $m = 5$ case. Here $X$ stands for any number of jets. We illustrate these signals with benchmark points consistent with the usual constraints like unitarity, perturbativity, the precision electroweak data, the observed dark matter relic density of the Universe and, most importantly, the stringent LHC constraints from the post Higgs ($h$) discovery era like the measured $M_h$ and the upper bound on the invisible width of $h$ decay which were not included in earlier analyses of multilepton signatures. We find that if the IDM model is embedded in a grand dessert scenario so that the unitarity constraint holds up to a very high scale, the whole of the highly restricted parameter space allowed by the above constraints can be probed at the LHC via the $3l$ signal for an integrated luminosity $\sim 3000 ~{\rm fb}^{-1}$. On the other hand, if any new physics shows up at a scale $\sim$ 10~TeV, only a part of the enlarged allowed parameter space can be probed. The $4l$ and $5l$ signals can help to discriminate among different IDM scenarios as and when sufficient integrated luminosity accumulates.

hep-ph↗

arXiv2016

New Limits on Heavier Electroweakinos and their LHC Signatures

We investigate the heavier electroweakino sectors in several versions of the MSSM, which has not been explored so far in the light of the LHC data, and obtain new bounds using the ATLAS Run I constraints in the $3l + {E\!\!\!\!/_T}$ channel. We also venture beyond the trilepton events and predict several novel multilepton + ${E\!\!\!\!/_T}$ signatures of these electroweakinos which may show up before the next shutdown of the LHC.

hep-ph↗

arXiv2016

Lepton flavor violating Higgs boson decay $h\toμτ$ at the ILC

We study the possible reach of the proposed International Linear Collider (ILC) in exploring the lepton flavor violating (LFV) Higgs boson decay $h\toμτ$. Two generic types of models are investigated, both involving an extended scalar sector. For the first type, the rest of the scalars are heavy and beyond the reach of ILC but the LFV decay occurs through a tiny admixture of the Standard Model (SM) doublet with the heavy degrees of freedom. In the second class, which is more constrained from the existing data, the SM Higgs boson does not have any LFV decay but there are other scalars degenerate with it that gives rise to the LFV signal. We show that ILC can pin down the branching fraction of $h\toμτ$, and hence the effective LFV Yukawa coupling, to a very small value, due to the fact that there are signal channels with unlike flavor leptons but no missing energy. It turns out that the low-energy options of the ILC, namely, $\sqrt{s}=250$ or 500 GeV are better to investigate such channels, and the option of beam polarization helps too. At least an order of magnitude improvement is envisaged over the existing limits.

hep-ph↗

arXiv2015

Reduced LHC constraints for higgsino-like heavier electroweakinos

As a sequel to our earlier work on wino-dominated $\tilde χ_1^{\pm}$ and $\tilde χ_2^{0}$ (wino models), we focus on the pMSSM models where $\tilde χ_1^{\pm}$ and $\tilde χ_{2,3}^{0}$ are either higgsino dominated (higgsino models) or admixtures of significant amount of higgsino and wino components (mixed models), with or without light sleptons. The LHC constraints in the trilepton channel are significantly weaker even in the presence of light sleptons, especially in the higgsino models, compared to those mostly studied by the LHC collaborations with wino-dominated $\tilde χ_1^{\pm}$ and $\tilde χ_2^{0}$. The modes $\tilde χ_{2,3}^{0}\rightarrow h~\tildeχ_1^{0}$ with large branching ratios (BRs) are more common in the higgsino models and may produce spectacular signal in the LHC Run-II. In a variety of higgsino and mixed models we have delineated the allowed parameter space due to the LHC constraints, the observed Dark Matter (DM) relic density of the universe, which gets contributions from many novel DM producing mechanisms i.e., the annihilation/coannihilation processes that lead to the correct range of relic density, and the precise measurement of the anomalous magnetic moment of the muon. In the higgsino models many new DM producing mechanisms, which are not allowed in the wino models, open up. We have also explored the prospects of direct and indirect detection of DM in the context of the LUX and IceCube experiments respectively. In an extended model having only light gluinos in addition to the electroweak sparticles, the gluinos decay into final states with multiple taggable b-jets with very large BRs. As a consequence, the existing ATLAS data in the $0l$ + jets (3b) + $E\!\!\!\!/_T$ channel provide the best limit on $m_{\tilde g}$ ($\approx$ 1.3 TeV). Several novel signatures of higgsino models for LHC Run-II and ILC have been identified.

hep-ph↗

arXiv2015

Gluino mass limits with sbottom NLSP in coannihilation scenarios

In view of the recent interest in the pMSSM with light third generation squarks, we consider a hitherto neglected scenario where the lighter bottom squark ($\widetilde{b}_1$) is the next lightest supersymmetric particle (NLSP) which co-annihilates with the lightest supersymmetric particle (LSP), the dark matter (DM) candidate. Since the co-annihilation cross section receives contributions from both electroweak and strong vertices, it is relatively large. As a result relatively large NLSP-LSP mass difference (25 - 35 GeV) is consistent with the PLANCK data. This facilitates the LHC signatures of this scenario. We consider several variants of the sbottom NLSP scenario with and without light stops and delineate the parameter space allowed by the PLANCK data. We point out several novel signal (e.g., $\widetilde{t}_1 \rightarrow \widetilde{b}_1 W$) which are not viable in the stop NLSP scenario of DM production. Finally, we consider gluino ($\widetilde g$) decays in this scenario and using the current ATLAS data in the jets (with or without b-tagging) + $\not \!\! E_T$ channel, obtain new limits in the $m_{\widetilde{b}_1} - m_{\widetilde g}$ mass plane. We find that for $m_{\widetilde{b}_1}$ upto 500 GeV, $m_{\widetilde g} \geq$ 1.1 - 1.2 TeV in this scenario.

hep-ph↗

arXiv2014

LHC signatures of neutrino mass generation through R-parity violation

We consider a model of neutrino mass based on R-parity violating (RPV) supersymmetry, with three $μ_i$, relevant for bilinear RPV terms, and three $λ'_{ijk}$, relevant for the trilinear terms. The present neutrino data, after a precise determination of the mixing angle $θ_{13}$, severely constrain such models. We make a thorough study of one such class of models that may have interesting signatures at the upgraded LHC. In this class of models, the relevant trilinear couplings are of the form $λ'_{i33}$, so if the lighter stop squark $\tilde{t}_1$ is the lightest supersymmetric particle (LSP), it will decay only through these couplings, giving rise to events with isolated hard leptons and jets. Even when $\tilde{t}_1$ is the next-to-LSP (NLSP), it can decay via the tiny $λ'$ couplings allowed by the neutrino data, although it may face stiff competition from some R-parity conserving decay modes. Using a simple Pythia based simulation, we find that in both the cases the signal consisting of a pair of oppositely charged leptons ($ee$, $μμ$ or $e μ$) plus jets may be observable at the upgraded LHC experiments for a reasonable range of the $\tilde{t}_1$ mass.

hep-ph↗

arXiv2014

The Electroweak Sector of the pMSSM in the Light of LHC - 8 TeV and Other Data

Using the chargino-neutralino and slepton search results from the LHC in conjunction with the WMAP/PLANCK and $(g-2)_μ$ data, we constrain several generic pMSSM models with decoupled strongly interacting sparticles, heavier Higgs bosons and characterized by different hierarchies among the EW sparticles. We find that some of them are already under pressure and this number increases if bounds from direct detection experiments like LUX are taken into account, keeping in mind the associated uncertainties. The XENON1T experiment is likely to scrutinize the remaining models closely. Analysing models with heavy squarks, a light gluino along with widely different EW sectors, we show that the limits on gluino mass are not likely to be below 1.1 TeV, if a multichannel analysis of the LHC data is performed. Using this light gluino scenario we further illustrate that in future LHC experiments the models with different EW sectors can be distinguished from each other by the relative sizes of the $n$-leptons + $m$-jets + ${\mbox{${E\!\!\!\!/_T}$}}$ signals for different choices of $n$.

hep-ph↗

arXiv2013

Neutralino dark matter confronted by the LHC constraints on Electroweak SUSY signals

The supersymmetric particles (sparticles) belonging exclusively to the electroweak sector of the minimal supersymmetric standard model (MSSM) may hold the key to the observed dark matter relic density in the universe even if all strongly interacting sparticles are very heavy. The importance of the light EW sparticles in DM physics and in producing spectacular collider signals is emphasized. It is shown that even the preliminary data on the direct searches of these sparticles at the LHC, significantly constrain the parameter space of the MSSM compatible with the observed relic density and provide useful hints about the future search prospects. If in addition to the electroweak sparticles the gluinos are also within the reach of the LHC experiments, then the gluino mass limits in the light slepton scenario obtained via the canonical jets + $\met$ channel may be relaxed by as much as 25 $%$ compared to the existing limits. But the corresponding same sign dilepton (SSD) + jets + $\met$ signal will yield enhanced limits competitive with the strongest limits currently available. This is illustrated with the help of benchmark scenarios at the generator level using PYTHIA. If the gluinos are just beyond the current reach of the LHC, then the generic n-lepton + m-jets + missing energy signal may discriminate between different DM producing mechanisms by comparing the signals corresponding to different values of n. This is illustrated by simulating the signals for n = 0 and n = 2 (the SSD signal).

hep-ph↗