Search

In the internet era, almost every business entity is trying to have its digital footprint in digital media and other social media platforms. For these entities, word of mouse is also very important. Particularly, this is quite crucial for the hospitality sector dealing with hotels, restaurants etc. Consumers do read other consumers reviews before making final decisions. This is where it becomes very important to understand which aspects are affecting most in the minds of the consumers while giving their ratings. The current study focuses on the consumer reviews of Indian hotels to extract aspects important for final ratings. The study involves gathering data using web scraping methods, analyzing the texts using Latent Dirichlet Allocation for topic extraction and sentiment analysis for aspect-specific sentiment mapping. Finally, it incorporates Random Forest to understand the importance of the aspects in predicting the final rating of a user., Comment: This is the pre-print of the paper that was accepted for oral presentation and publication in the proceedings of IEEE ICCCNT 2024 which was organized as IIT Mandi, India from June 24 to 28, 2024. The paper is 5 pages long and it contains 4 figures and 6 tables. The is not the final version of the paper

The Mg II k \& h line intensity ratios can be used to probe the characteristics of the plasma in the solar atmosphere. In this study, using the observations recorded by the Interface Region Imaging Spectrometer (IRIS), we study the variation of the Mg II k \& h intensity ratio for three flares belonging to X-class, M-class, and C-class, throughout their evolution. We also study the k-to-h intensity ratio as a function of magnetic flux density obtained from the line-of-sight magnetograms recorded by the Helioseismic and Magnetic Imager (HMI) on board the Solar Dynamics Observatory (SDO). Our results reveal that while the intensity ratios are independent of magnetic flux density, they show significant changes during the evolution of the C-class and M-class flares. The intensity ratios start to increase at the start of the flare and peak during the impulsive phase before the flare peak and decrease rapidly thereafter. The values of the ratios fall even below the pre-flare level during the peak and decline phases of the flare. These results are important in light of the heating and cooling of localized plasma and provide further constraints on the understanding of flare physics., Comment: Accepted for publication in ApJ

This paper proposes a simple but highly efficient expansion-based model for continual learning. The recent feature transformation, masking and factorization-based methods are efficient, but they grow the model only over the global or shared parameter. Therefore, these approaches do not fully utilize the previously learned information because the same task-specific parameter forgets the earlier knowledge. Thus, these approaches show limited transfer learning ability. Moreover, most of these models have constant parameter growth for all tasks, irrespective of the task complexity. Our work proposes a simple filter and channel expansion based method that grows the model over the previous task parameters and not just over the global parameter. Therefore, it fully utilizes all the previously learned information without forgetting, which results in better knowledge transfer. The growth rate in our proposed model is a function of task complexity; therefore for a simple task, the model has a smaller parameter growth while for complex tasks, the model requires more parameters to adapt to the current task. Recent expansion based models show promising results for task incremental learning (TIL). However, for class incremental learning (CIL), prediction of task id is a crucial challenge; hence, their results degrade rapidly as the number of tasks increase. In this work, we propose a robust task prediction method that leverages entropy weighted data augmentations and the models gradient using pseudo labels. We evaluate our model on various datasets and architectures in the TIL, CIL and generative continual learning settings. The proposed approach shows state-of-the-art results in all these settings. Our extensive ablation studies show the efficacy of the proposed components., Comment: To be Appeared in WACV, 2024

The thermal Sunyaev-Zeldovich (SZ) effect serves as a direct potential probe of the energetic outflows from quasars that are responsible for heating the intergalactic medium. In this work, we use the GIZMO meshless finite mass hydrodynamic cosmological simulation SIMBA (Dave et al. 2019), which includes different prescriptions for quasar feedback, to compute the SZ effect arising from different feedback modes. From these theoretical simulations, we perform mock observations of the Atacama Large Millimeter Array (ALMA) in four bands (320 GHz, 135 GHZ, 100 GHz and 42 GHz) to characterize the feasibility of direct detection of the quasar SZ signal. Our results show that for all the systems we get an enhancement of the SZ signal, when there is radiative feedback, while the signal gets suppressed when the jet mode of feedback is introduced in the simulations. Our mock ALMA maps reveal that, with the current prescription of jet feedback, the signal goes below the detection threshold of ALMA. We also find that the signal is higher for high redshift systems, making it possible for ALMA and cross SZ-X-ray studies to disentangle the varying modes of quasar feedback and their relative importance in the cosmological context., Comment: Accepted for publication in ApJ

To investigate the effect of feedback from active galactic nuclei (AGN) on their surrounding medium, we study the diffuse X-ray emission from galaxy groups and clusters by coupling the Astrophysical Plasma Emission Code (APEC) with the cosmological hydrodynamic simulation involving AGN feedback. We construct a statistical sample of synthetic Chandra X-ray photon maps to observationally characterize the effect of AGN on the ambient medium. We show that AGN are effective in displacing the hot X-ray emitting gas from the centers of groups and clusters, and that these signatures remain evident in observations of the X-ray surface brightness profiles., Comment: Accepted for publication in Astronomische Nachrichten, proceedings of the 9th International Workshop on Astronomy and Relativistic Astrophysics

Given an image, we would like to learn to detect objects belonging to particular object categories. Common object detection methods train on large annotated datasets which are annotated in terms of bounding boxes that contain the object of interest. Previous works on object detection model the problem as a structured regression problem which ranks the correct bounding boxes more than the background ones. In this paper we develop algorithms which actively obtain annotations from human annotators for a small set of images, instead of all images, thereby reducing the annotation effort. Towards this goal, we make the following contributions: 1. We develop a principled version space based active learning method that solves for object detection as a structured prediction problem in a weakly supervised setting 2. We also propose two variants of the margin sampling strategy 3. We analyse the results on standard object detection benchmarks that show that with only 20% of the data we can obtain more than 95% of the localization accuracy of full supervision. Our methods outperform random sampling and the classical uncertainty-based active learning algorithms like entropy

Nuclear Magnetic Resonance (NMR) forms a natural test-bed to perform quantum information processing (QIP) and has so far proven to be one of the most successful quantum information processors. The nuclear spins in a molecule treated as quantum bits or qubits which are the basic building blocks of a quantum computer. The development of NMR over half a century puts it in a platform where we can utilize its excellent control techniques over an ensemble of spin systems and perform quantum computation in a highly controlled way. Apart from a successful quantum information processor, NMR is also a highly powerful quantum platform where many of the potentially challenging quantum mechanical experiments can be performed., Comment: PhD Thesis, 193 pages, 49 figures

We investigate evolution of quantum correlations in ensembles of two-qubit nuclear spin systems via nuclear magnetic resonance techniques. We use discord as a measure of quantum correlations and the Werner state as an explicit example. We first introduce different ways of measuring discord and geometric discord in two-qubit systems and then describe the following experimental studies: (a) We quantitatively measure discord for Werner-like states prepared using an entangling pulse sequence. An initial thermal state with zero discord is gradually and periodically transformed into a mixed state with maximum discord. The experimental and simulated behavior of rise and fall of discord agree fairly well. (b) We examine the efficiency of dynamical decoupling sequences in preserving quantum correlations. In our experimental setup, the dynamical decoupling sequences preserved the traceless parts of the density matrices at high fidelity. But they could not maintain the purity of the quantum states and so were unable to keep the discord from decaying. (c) We observe the evolution of discord for a singlet-triplet mixed state during a radio-frequency spin-lock. A simple relaxation model describes the evolution of discord, and the accompanying evolution of fidelity of the long-lived singlet state, reasonably well., Comment: 9 pages, 7 figures, Phys. Rev. A (in press)

We report the first experimental demonstration of quantum delayed-choice experiment via nuclear magnetic resonance techniques. An ensemble of molecules each with two spin-1/2 nuclei are used as target and the ancilla qubits to perform the quantum circuit corresponding the delayed-choice setup. As expected in theory, our experiments clearly demonstrate the continuous morphing of the target qubit between particle-like and wave-like behaviors. The experimental visibility of the interference patterns shows good agreement with the theory., Comment: Revised text, more figures added

The long-lived singlet states are useful to study a variety of interesting quantum phenomena. In this work we study electromagnetically induced transparency using a two-qubit system. The singlet state acts as a `dark state' which does not absorb a probe radiation in the presence of a control radiation. Further we demonstrate that the simultaneous irradiation of probe and control radiations acts as a dynamical decoupling preserving the singlet state at higher correlation for longer durations., Comment: 4 pages, 4 figures