arXiv subjects

Sneha Chaudhari

Publications and source records attributed to Sneha Chaudhari.

4 recordsLinked to original sources

arXiv2021

An Attentive Survey of Attention Models

Attention Model has now become an important concept in neural networks that has been researched within diverse application domains. This survey provides a structured and comprehensive overview of the developments in modeling attention. In particular, we propose a taxonomy which groups existing techniques into coherent categories. We review salient neural architectures in which attention has been incorporated, and discuss applications in which modeling attention has shown a significant impact. We also describe how attention has been used to improve the interpretability of neural networks. Finally, we discuss some future research directions in attention. We hope this survey will provide a succinct introduction to attention models and guide practitioners while developing approaches for their applications.

cs.LG↗

arXiv2021

Spotless days and geomagnetic index as the predictors of solar cycle 25

We study the sunspot activity in relation to spotless days (SLDs) during the descending phase of solar cycle $11$--$24$ to predict the amplitude of sunspot cycle $25$. For this purpose, in addition to SLD, we also use the geomagnetic activity (aa index) during the descending phase of a given cycle. A very strong correlation of the SLD (R=$0.68$) and aa index (R=$0.86$) during the descending phase of a given cycle with the maximum amplitude of next solar cycle has been estimated.The empirical relationship led us to deduce the amplitude of cycle $25$ to be 99.13$\pm$ 14.97 and 104.23$\pm$ 17.35 using SLD and aa index, respectively as predictors.Both the predictors provide comparable amplitude for solar cycle $25$ and reveal that the solar cycle $25$ will be weaker than cycle $24$. Further we derive that the maximum of cycle $25$ is likely to occur between February and March 2024. While the aa index has been used extensively in the past, this work establishes SLDs as another potential candidate for predicting the characteristics of the next cycle.

astro-ph.SR↗

arXiv2018

Study of temporal and spectral characteristics of the X-ray emission from solar flares

Temporal and spectral characteristics of X-ray emission from 60 flares of intensity $\ge$C class observed by Solar X-ray Spectrometer (SOXS) during 2003-2011 are presented. We analyse the X-ray emission observed in four and three energy bands by the Si and CZT detectors, respectively. The number of peaks in the intensity profile of the flares varies between 1 and 3. We find moderate correlation (R$\simeq$0.2) between the rise time and the peak flux of the first peak of the flare irrespective to energy band, which is indicative of its energy-independent nature. Moreover, magnetic field complexity of the flaring region is found to be anti-correlated (R=0.61) with the rise time of the flares while positively correlated (R=0.28) with the peak flux of the flare. The time delay between the peak of the X-ray emission in a given energy band and that in the 25-30 keV decreases with increasing energy suggesting conduction cooling to be dominant in the lower energies. Analysis of 340 spectra from 14 flares reveals that the peak of Differential Emission Measure (DEM) evolution delays by 60-360 s relative to that of the temperature, and this time delay is inversely proportional to the peak flux of the flare. We conclude that temporal and intensity characteristics of flares are energy dependent as well as magnetic field configuration of the active region.

astro-ph.SR↗

arXiv2014

Test Set Selection using Active Information Acquisition for Predictive Models

In this paper, we consider active information acquisition when the prediction model is meant to be applied on a targeted subset of the population. The goal is to label a pre-specified fraction of customers in the target or test set by iteratively querying for information from the non-target or training set. The number of queries is limited by an overall budget. Arising in the context of two rather disparate applications- banking and medical diagnosis, we pose the active information acquisition problem as a constrained optimization problem. We propose two greedy iterative algorithms for solving the above problem. We conduct experiments with synthetic data and compare results of our proposed algorithms with few other baseline approaches. The experimental results show that our proposed approaches perform better than the baseline schemes.

cs.AI↗