arXiv subjects

Gaurav Singh

Publications and source records attributed to Gaurav Singh.

At least 55 records · Page 3Linked to original sources

arXiv2021

Sentiment Analysis of Code-Mixed Social Media Text (Hinglish)

This paper discusses the results obtained for different techniques applied for performing the sentiment analysis of social media (Twitter) code-mixed text written in Hinglish. The various stages involved in performing the sentiment analysis were data consolidation, data cleaning, data transformation and modelling. Various data cleaning techniques were applied, data was cleaned in five iterations and the results of experiments conducted were noted after each iteration. Data was transformed using count vectorizer, one hot vectorizer, tf-idf vectorizer, doc2vec, word2vec and fasttext embeddings. The models were created using various machine learning algorithms such as SVM, KNN, Decision Trees, Random Forests, Naive Bayes, Logistic Regression, and ensemble voting classifiers. The data was obtained from a task on Codalab competition website which was listed as Task:9 on the Semeval-2020 competition website. The models created were evaluated using the F1-score (macro). The best F1-score of 69.07 was achieved using ensemble voting classifier.

cs.CL↗

arXiv2021

Study of dynamical status of the globular cluster NGC 1851 using Ultraviolet Imaging Telescope

We present the study of dynamical status of the globular cluster NGC 1851. A combination of multiwavelength space and ground-based data sets are used for the present analysis. In order to select the genuine cluster members, we used the astro-photometric data available from HST and GAIA-DR2 catalogs. The BSS radial distribution of the cluster is plotted from the center of the cluster to the outskirts. The radial distribution of BSS shows a central peak, followed by a dip at the intermediate radii (rmin ~ 90'') and a rising trend in the outskirts. We also estimated A+rh parameter as 0.391 +/- 0.006 to validate the findings of the radial distribution study. On the basis of the minima in the BSS radial distribution and the value of A+rh parameter, we conclude that NGC 1851 belongs to Family II classification and is an intermediate dynamical state cluster.

astro-ph.GA↗

arXiv2020

Simulating multiscale gated field emitters -- a hybrid approach

Multi-stage cathodes are promising candidates for field emission due to the multiplicative effect in local field predicted by the Schottky conjecture and its recent corrected counterpart [J. Vac. Sci. Technol. B 38, 023208 (2020)]. Due to the large variation in length scales even in a 2-stage compound structure consisting of a macroscopic base and a microscopic protrusion, the simulation methodology of a gated field emitting compound diode needs to be revisited. As part of this strategy, the authors investigate the variation of local field on the surface of a compound emitter near its apex and find that the generalized cosine law continues to hold locally near the tip of a multi-scale gated cathode. This is used to emit charges with appropriate distributions in position and velocity components with a knowledge of only the electric field at the apex. The distributions are consistent with contemporary free-electron field emission model and follow from the joint distribution of launch angle, total energy, and normal energy. For a compound geometry with local field enhancement by a factor of around 1000, a hybrid model is used where the vacuum field calculated using COMSOL is imported into the Particle-In-Cell code PASUPAT where the emission module is implemented. Space charge effects are incorporated in a multi-scale adaptation of PASUPAT using a truncated geometry with `open electrostatic boundary' condition. The space charge field, combined with the vacuum field, is used for particle-emission and tracking.

physics.app-ph↗

arXiv2020

Peculiarities in the horizontal branch stars of globular cluster NGC 1851: Discovery of a Blue straggler companion to an EHB star

We present a study of Far Ultraviolet (FUV) bright horizontal branch (HB) stars to understand the peculiarities seen in the HB sequence of the Globular Cluster NGC 1851, using ground and space based multi-wavelength data. Optical and UV color magnitude diagrams are used to classify HB stars and their membership from HST and Gaia DR2 data. The Spectral energy distributions (SEDs) of the hot HB stars located from the core to tidal radii are constructed. The SEDs reveal that the HB stars near the G-jump show a decrease in the FUV flux when atmospheric models of cluster metallicity are used for fitting, but a better fit is found with higher metallicity models, as expected due to atmospheric diffusion. We report on four particularly interesting extreme HB (EHB) stars, two each in inner and outer regions. We detect a sub-luminous EHB and a "blue-Hook" candidates with temperatures Teff ~ 25,000 K and 31,000 K, respectively. We found an EHB star (Teff ~ 17,000 K) with the radius lies between the BHB and normal EHB stars. The most peculiar of our EHB stars (Teff ~ 28,000 K) is found to be a photometric binary to a Blue Straggler star (BSS) (Teff ~ 7,000 K), which is an important target for spectroscopic study. This discovery of the candidate EHB+BSS binary system, could help to explain the mass loss in the RGB phase, leading to the formation of EHB stars.

astro-ph.SR↗

arXiv2020

Decision Tree J48 at SemEval-2020 Task 9: Sentiment Analysis for Code-Mixed Social Media Text (Hinglish)

This paper discusses the design of the system used for providing a solution for the problem given at SemEval-2020 Task 9 where sentiment analysis of code-mixed language Hindi and English needed to be performed. This system uses Weka as a tool for providing the classifier for the classification of tweets and python is used for loading the data from the files provided and cleaning it. Only part of the training data was provided to the system for classifying the tweets in the test data set on which evaluation of the system was done. The system performance was assessed using the official competition evaluation metric F1-score. Classifier was trained on two sets of training data which resulted in F1 scores of 0.4972 and 0.5316.

cs.CL↗

arXiv2020

Neighborhood Sensitive Mapping for Zero-Shot Classification using Independently Learned Semantic Embeddings

In a traditional setting, classifiers are trained to approximate a target function $f:X \rightarrow Y$ where at least a sample for each $y \in Y$ is presented to the training algorithm. In a zero-shot setting we have a subset of the labels $\hat{Y} \subset Y$ for which we do not observe any corresponding training instance. Still, the function $f$ that we train must be able to correctly assign labels also on $\hat{Y}$. In practice, zero-shot problems are very important especially when the label set is large and the cost of editorially label samples for all possible values in the label set might be prohibitively high. Most recent approaches to zero-shot learning are based on finding and exploiting relationships between labels using semantic embeddings. We show in this paper that semantic embeddings, despite being very good at capturing relationships between labels, are not very good at capturing the relationships among labels in a data-dependent manner. For this reason, we propose a novel two-step process for learning a zero-shot classifier. In the first step, we learn what we call a \emph{property embedding space} capturing the "\emph{learnable}" features of the label set. Then, we exploit the learned properties in order to reduce the generalization error for a linear nearest neighbor-based classifier.

cs.LG↗

arXiv2020

Enhanced space charge limited current for curved electron emitters

The maximum current that can be transported across a vacuum diode is limited by forces arising due to space charge. In a planar diode configuration, the space charge limited (SCL) current density from a planar emitting patch is given by the Child-Langmuir (CL) law $J_{CL} \sim V_g^{3/2}/D^2$ where $V_g$ is the potential difference across the diode and $D$ is the separation between the anode and cathode. We show here analytically using the nonlinear line charge model that for a curved emitter in a planar diode configuration, the limiting current obeys the scaling relationship $J_{SCL} \sim γ_a V_g^{3/2}/D^2$ where $γ_a$ is the apex field enhancement factor of the curved emitter. For an emitter with large height ($h$) to apex radius of curvature ($R_a$) ratio, the limiting current far exceeds the planar value. The result is verified using the particle-in-cell code PASUPAT for two curved emitters shapes.

physics.plasm-ph↗

arXiv2020

Constructing Artificial Data for Fine-tuning for Low-Resource Biomedical Text Tagging with Applications in PICO Annotation

Biomedical text tagging systems are plagued by the dearth of labeled training data. There have been recent attempts at using pre-trained encoders to deal with this issue. Pre-trained encoder provides representation of the input text which is then fed to task-specific layers for classification. The entire network is fine-tuned on the labeled data from the target task. Unfortunately, a low-resource biomedical task often has too few labeled instances for satisfactory fine-tuning. Also, if the label space is large, it contains few or no labeled instances for majority of the labels. Most biomedical tagging systems treat labels as indexes, ignoring the fact that these labels are often concepts expressed in natural language e.g. `Appearance of lesion on brain imaging'. To address these issues, we propose constructing extra labeled instances using label-text (i.e. label's name) as input for the corresponding label-index (i.e. label's index). In fact, we propose a number of strategies for manufacturing multiple artificial labeled instances from a single label. The network is then fine-tuned on a combination of real and these newly constructed artificial labeled instances. We evaluate the proposed approach on an important low-resource biomedical task called \textit{PICO annotation}, which requires tagging raw text describing clinical trials with labels corresponding to different aspects of the trial i.e. PICO (Population, Intervention/Control, Outcome) characteristics of the trial. Our empirical results show that the proposed method achieves a new state-of-the-art performance for PICO annotation with very significant improvements over competitive baselines.

cs.CL↗

arXiv2019

Relation Extraction using Explicit Context Conditioning

Relation Extraction (RE) aims to label relations between groups of marked entities in raw text. Most current RE models learn context-aware representations of the target entities that are then used to establish relation between them. This works well for intra-sentence RE and we call them first-order relations. However, this methodology can sometimes fail to capture complex and long dependencies. To address this, we hypothesize that at times two target entities can be explicitly connected via a context token. We refer to such indirect relations as second-order relations and describe an efficient implementation for computing them. These second-order relation scores are then combined with first-order relation scores. Our empirical results show that the proposed method leads to state-of-the-art performance over two biomedical datasets.

cs.CL↗

arXiv2018

The radial distribution of Blue stragglers in Galactic Globular Cluster NGC 6656 $-$ Clues on the dynamical status

We present dynamical status of the Galactic globular cluster NGC 6656 using spatial distribution of Blue Straggler Stars (BSSs). A combination of multi-wavelength high-resolution space and ground-based data are used to cover a large cluster region. We determine the centre of gravity ($C_{grav}$) and construct the projected density profile of the cluster using the probable cluster members selected from HST and Gaia DR2 proper motion data sets. The projected density profile in the investigated region is nicely reproduced by a single mass King model, with core ($r_{c}$) and tidal ($r_{t}$) radius as $75^{\prime\prime}.2$ $\pm$ $3^{\prime\prime}.1$ and $35^{\prime}.6$ $\pm$ $1^{\prime}.1$ respectively. In total, 90 BSSs are identified on the basis of proper motion data in the region of radius $625^{\prime\prime}$. An average mass of the BSSs is determined as $1.06$ $\pm$ $0.09$ $M_{\odot}$ and with an age range of 0.5 to 7 Gyrs. The BSS radial distribution shows a bimodal trend, with a peak in the centre, a minimum at $r \sim r_c$ and a rising tendency in the outer region. The BSS radial distribution shows a flat behaviour in the outermost region of the cluster. We also estimate $A^{+}_{rh}$ parameter as an alternative indicator of the dynamical status of the cluster and is found to be $0.038$ $\pm$ $0.016$. Based on the radial distribution and $A^{+}_{rh}$ parameter, we conclude that NGC 6656 is an intermediate dynamical age cluster.

astro-ph.SR↗

arXiv2018

The cosine law of field enhancement factor variation: generic emitter shapes

The cosine law of field enhancement factor variation was recently derived for a hemi-ellipsoidal emitter and numerically established for other smooth emitter shapes (Biswas et al, Ultramicroscopy, 185, 1 (2018)). An analytical derivation is provided here for general smooth vertical emitter shapes aligned in the direction of the asymptotic electrostatic field. The law is found to hold in the neighbourhood of the emitter apex from where field emisson pre-dominantly occurs.

physics.app-ph↗

arXiv2018

Faster Convergence & Generalization in DNNs

Deep neural networks have gained tremendous popularity in last few years. They have been applied for the task of classification in almost every domain. Despite the success, deep networks can be incredibly slow to train for even moderate sized models on sufficiently large datasets. Additionally, these networks require large amounts of data to be able to generalize. The importance of speeding up convergence, and generalization in deep networks can not be overstated. In this work, we develop an optimization algorithm based on generalized-optimal updates derived from minibatches that lead to faster convergence. Towards the end, we demonstrate on two benchmark datasets that the proposed method achieves two orders of magnitude speed up over traditional back-propagation, and is more robust to noise/over-fitting.

cs.LG↗

arXiv2018

Structured Multi-Label Biomedical Text Tagging via Attentive Neural Tree Decoding

We propose a model for tagging unstructured texts with an arbitrary number of terms drawn from a tree-structured vocabulary (i.e., an ontology). We treat this as a special case of sequence-to-sequence learning in which the decoder begins at the root node of an ontological tree and recursively elects to expand child nodes as a function of the input text, the current node, and the latent decoder state. In our experiments the proposed method outperforms state-of-the-art approaches on the important task of automatically assigning MeSH terms to biomedical abstracts.

cs.IR↗

arXiv2018

Improving Active Learning in Systematic Reviews

Systematic reviews are essential to summarizing the results of different clinical and social science studies. The first step in a systematic review task is to identify all the studies relevant to the review. The task of identifying relevant studies for a given systematic review is usually performed manually, and as a result, involves substantial amounts of expensive human resource. Lately, there have been some attempts to reduce this manual effort using active learning. In this work, we build upon some such existing techniques, and validate by experimenting on a larger and comprehensive dataset than has been attempted until now. Our experiments provide insights on the use of different feature extraction models for different disciplines. More importantly, we identify that a naive active learning based screening process is biased in favour of selecting similar documents. We aimed to improve the performance of the screening process using a novel active learning algorithm with success. Additionally, we propose a mechanism to choose the best feature extraction method for a given review.

cs.IR↗

arXiv2017

The tunneling potential for field emission from nanotips

In the quasi-planar approximation of field emission, the potential energy due to an external electrostatic field $E_0$ is expressed as $-e γE_0 Δs$ where $Δs$ is the perpendicular distance from the emission site and $γ$ is the local field enhancement factor on the surface of the emitter. We show that for curved emitter tips, the current density can be accurately computed if terms involving $(Δs/R_2)^2$ and $(Δs/R_2)^3$ are incorporated in the potential where $R_2$ is the second (smaller) principle radius of curvature. The result is established analytically for the hemiellipsoid and hyperboloid emitters and it is found that for sharply curved emitters, the expansion coefficients are equal and coincide with that of a sphere. The expansion seems to be applicable to generic emitters as demonstrated numerically for an emitter with a conical base and quadratic tip. The correction terms in the potential are adequate for $R_a \gtrapprox 2$ nm for local field strengths of $5$ V/nm or higher. The result can also be used for nano-tipped emitter arrays or even a randomly placed bunch of sharp emitters.

cond-mat.mes-hall↗

arXiv2017

Variation of field enhancement factor near the emitter tip

The field enhancement factor at the emitter tip and its variation in a close neighbourhood determines the emitter current in a Fowler-Nordheim like formulation. For an axially symmetric emitter with a smooth tip, it is shown that the variation can be accounted by a $\cos{\tildeθ}$ factor in appropriately defined normalized co-ordinates. This is shown analytically for a hemi-ellipsoidal emitter and confirmed numerically for other emitter shapes with locally quadratic tips.

physics.app-ph↗

arXiv2016

Predicting Branch Visits and Credit Card Up-selling using Temporal Banking Data

There is an abundance of temporal and non-temporal data in banking (and other industries), but such temporal activity data can not be used directly with classical machine learning models. In this work, we perform extensive feature extraction from the temporal user activity data in an attempt to predict user visits to different branches and credit card up-selling utilizing user information and the corresponding activity data, as part of \emph{ECML/PKDD Discovery Challenge 2016 on Bank Card Usage Analysis}. Our solution ranked \nth{4} for \emph{Task 1} and achieved an AUC of \textbf{$0.7056$} for \emph{Task 2} on public leaderboard.

cs.LG↗

arXiv2016

Efficient Document Indexing Using Pivot Tree

We present a novel method for efficiently searching top-k neighbors for documents represented in high dimensional space of terms based on the cosine similarity. Mostly, documents are stored as bag-of-words tf-idf representation. One of the most used ways of computing similarity between a pair of documents is cosine similarity between the vector representations, but cosine similarity is not a metric distance measure as it doesn't follow triangle inequality, therefore most metric searching methods can not be applied directly. We propose an efficient method for indexing documents using a pivot tree that leads to efficient retrieval. We also study the relation between precision and efficiency for the proposed method and compare it with a state of the art in the area of document searching based on inner product.

cs.IR↗