ISI Digital Commons (Indian Statistical Institute )
Not a member yet
7571 research outputs found
Sort by
A new multimodal sentiment analysis for images containing textual information
No full textMultimodal sentiment analysis on images with textual content is a research area aiming to understand the sentiment conveyed by visual and textual elements in the images. While multimodal sentiment analysis on images and text (reviews) has its own challenges, the combination of textual and visual content in the form of images presents new challenges as well as opportunities. In this research work, we proposed a multimodal sentiment analysis method that works on images incorporating textual elements. In the textual sentiment analysis model, we initially employed a recognition system to extract textual data from input images. Our proposed multimodal method is based on transfer learning, considering two pre-trained deep learning models, Xception, and RoBERTa, to extract features from both visual and textual content from multimedia images. We then implemented a fusion strategy to combine these two modalities (Visual Sentiment Analysis (VSA) and Textual Sentiment Analysis (TSA)) to enhance the accuracy of the proposed method and to provide a more comprehensive understanding of sentiment in multimedia content. In addition, we curated a custom dataset comprising images with associated text labels and sentiments. To ensure accurate labels, we conducted human evaluations involving thirty annotators. Our dataset includes images labeled with negative, neutral, and positive sentiments. Experimental results demonstrated the effectiveness of combining visual and textual features for sentiment analysis. The findings from this research hold promising implications for real-world applications, such as sentiment analysis in social media, product reviews, and marketing campaigns, where both images and text play a significant role in conveying emotional context
Essays on Choice and Matching
No full textThis thesis consists of three independent essays. The first chapter introduces a model of decision-making that is based on the procedure of rejection. Departing from the standard model of choice via preference maximization, the decision maker (DM) rejects minimal alternatives from a menu according to a preference relation. We axiomatically study the correspondence of non-rejected alternatives which we call the acceptable correspondence with different rationality conditions on the underlying preference relation. We also gen- eralize our model to acceptable correspondences that are generated by the successive elimination of minimal alternatives. We find that the rejection approach developed in this chapter can offer explanations for various anomalies observed in decision theory, such as the two-decoy effect or the two-compromise effect (Tserenjigmid (2019)). The second chapter proposes a sequential model of the college admissions problem. The selection criteria of institutions are formulated via choice rules that admit slot- specific priorities introduced by Kominers and S¨onmez (2016). We show that the appli- cants can not be worse off in the subsequent stages when the candidates update their preferences that adhere to their assignment in the previous stage. Moreover, the mech- anism that sequentially implements individual-proposing deferred acceptance is stable with respect to a generalized version of a sequential stability notion provided in this chapter. These results generalize the findings presented in Haeringer and Iehl´e (2021). We use our results to analyze recently reformed admission procedures for engineering colleges in India (Baswana et al. (2019)), where applicants are provided various options to update their preferences in additional stages. In the third chapter, we study the welfare consequences of merging Shapley–Scarf housing markets (Shapley and Scarf (1974)). We show that in the worst-case scenario, market integration can lead to large welfare losses and make the vast majority of agents worse off. However, on average, the integration is welfare enhancing and makes all agents better off ex-ante. The number of agents harmed by integration is a minority when all markets are small or the agent’s preferences are highly correlated
Improved lower and upper bounds on the span of distance labeling for some infinite regular grids
No full textThe channel assignment problem, popularly known as CAP, is one of the elementary and much studied topic in the field of wireless communication. The basic purpose for studying CAP is to find out solutions such that wireless communication becomes interference free with using spectrum as less as possible during the communication. Often the CAP is modeled as an L(k1, . . . , kℓ)-vertex (edge) labeling problem of a graph, where k1, . . . , kℓ are non-negative integers. In L(k1, . . . , kℓ)-vertex (edge) labeling problem, labels are assigned to the vertices (edges) of a graph in such a way that the absolute difference between the labels assigned to any pair of vertices (edges) located at distance i, 1 ≤ i ≤ ℓ, is ki. One of the objective of L(k1, . . . , kℓ)- vertex (edge) labeling of a graph G is to find a labeling of the vertices (edges) such that the span for the corresponding labeling is minimum among all L(k1, . . . , kℓ)- vertex (edge) labelings of G, where span denotes the difference between maximum and minimum labels used for a labeling. Regular grid graphs are common choices for modeling CAP because of their natural resemblance to cellular network for regular geometric pattern. Consequently, various studies of L(k1, k2, . . . , kℓ)-vertex (edge) labeling have been done for infinite regular grids such as infinite hexagonal (T3), square (T4), triangular (T6) and infinite 8-regular grid (T8) grids. In this thesis, we first derive the exact values of the span of L(1, 2)-edge labeling problem for T3 and T4. Then we improve the lower bound on the span of L(1, 2)-edge labeling problem for T6. Next by improving the lower bound, we derive the exact value of the span of L(1, 2)-edge labeling of T8. Next we attempt to derive theoretically the lower bound on the span of L(k1, k2)-vertex labeling problem for T6 for k1 ≤ k2. For this problem, the previous results were obtained partially through computer simulations. We find that our theoretically obtained results exactly coincide with the known results for the sub interval 0 ≤ k1 k2 ≤ 1 3 but provide loose bound for the other sub interval 1 3 ≤ k1 k2 ≤ 1. Next we derive improved lower bound on the span of L(2, 1)-edge labeling problem for T6. Next we study the L(1|, 1,{.z. . , 1} ℓ )-vertex labeling problem for T3. The exact value of the span of L(1|, 1,{.z. . , 1} ℓ )-vertex labeling problem for T3 has not been determined yet for any even ℓ ≥ 8, rather the value of the span was conjectured. We prove this conjecture for ℓ ≥ 8. In all the cases we analyze the structural properties of the underlined graphs and based on which the results are obtained theoreticall
A novel approach for retrieving GPP of evergreen forest regions of India using random forest regression
No full textGross Primary Productivity (GPP) is a crucial variable of global carbon cycle for determining the ecosystem\u27s health. Various methods are devised to quantify GPP and upscale it in both time and space. The most common methods are physical model and eddy covariance-based estimation, which are very restricted to surrounding area of study, only. The alternative methods are empirical (e.g., LUE, CASA, SCARF, and MODIS) and Machine learning (ML) models that employ remote sensing satellite data and geographical factors. However, for using ML models, ground-based measurements of GPP is a very important factor, which is not available in most places. We propose an alternative and effective way of estimating the GPP using the ML model and data from various flux sites around the globe for a particular plant functional type (PFT). In the present study, RF is used as ML model, which is trained on global GPP data from evergreen forest and implemented in Indian region. The key findings indicated that ML-based GPP is highly accurate and hence, we generated 20 years of time series GPP dataset (2001–2020). We validated with ground-based flux tower observations during 2016–2018 for three sites (very limited datasets) in India and compared them with MODIS GPP. The coefficient of determination (R2) value of the ML-based model was 0.84 with root mean square error (RMSE) of 1.45 gC m−2 Day−1 and mean absolute error (MAE) of 0.838 gC m−2 Day−1. The proposed approach is highly accurate and far better than the MODIS-based GPP. Therefore, it can be further extended to other forest types for a holistic assessment of the carbon cycle of a region
A novel attempt to describe the impact of infectious disease on the nation’s economy: an illustration through the Econo-epidemics model
No full textThe impacts of pandemic situation on the health and economy of any nation are indeed significant. Several paradigms are present, where the nations’s economy has stumbled due to impact of infectious diseases. Hence, we formulate an econo-epidemics model to study economic impact of infectious disease. The structure is amalgamated with Solow growth functions to describe effect of epidemiological factors on economy. Numerous numerical experiments, viz. bifurcation analysis, production of capital growth profile, etc., is conducted to provide valuable insights into the potential outcomes and dynamics of the economy. Additionally, we consider the impact of demographic stochasticity on the economy of third-tier countries, Bangladesh and Tanzania, as these nations face unique challenges and vulnerabilities. Enumeration of probability of economic losses in upcoming years and stationary distributions on the capital growth data of those countries provide an outline of strengthening the economy after a period of depreciation caused by an infectious disease outbreak. Overall, the statistical experiments and analysis of capital data provide valuable information about the potential economic consequences of infectious diseases. This knowledge can guide the formulation of policies and schemes to mitigate economic disasters in the future and promote a resilient economy that is better prepared to handle pandemics
A solar investigation of multicomponent dark matter
No full textIf multiple thermal weakly interacting massive particle (WIMP) dark matter candidates exist, then their capture and annihilation dynamics inside a massive star such as Sun could change from conventional method of study. With a simple correction to time evolution of dark matter (DM) number abundance inside the Sun for multiple dark matter candidates, significant changes in DM annihilation flux depending on annihilation, direct detection cross-section, internal conversion and their contribution to relic abundance are reported in present work
A solvable two-dimensional swarmalator model
No full textSwarmalators are oscillators that swarm through space as they synchronize in time. Introduced a few years ago to model many systems that mix synchrony with self-assembly, they remain poorly understood theoretically. Here, we obtain the first analytic results on swarmalators moving in two spatial dimensions by introducing a simplified model where the swarmalators have no hard-shell interaction terms and move on a periodic plane. These simplifications allow expressions for order parameters, stabilities and bifurcations to be derived exactly
Adjustments of flower opening time and duration in tropical rice (Oryza sativa ssp. indica) landraces in response to heat stress
No full textBased on our primary database of the flower opening time (FOT) and flower exposure duration (FED) of 1114 rice (Oryza sativa ssp. indica) landraces, we examined the influences of growing season, sunrise time as well as day maximum and minimum temperatures on the anthesis behaviour of indica rice landraces of South and Southeast Asia, flowering in summer and winter in 3 consecutive years (2020–2022). We also compared the FOT and FED on sunny and cloudy days of a small set of landraces, and also during summer and winter. Our data show that rice florets tend to open later in the morning and lengthen the sunrise-to-anthesis duration (SAD) on hotter sunny days during tropical summer than during winter and on cloudy days. These findings contradict the widely held conjecture, based on studies conducted at colder latitudes, that rice flowers open earlier in the morning to avoid heat stress. We propose that indica rice landraces are sufficiently adapted to tropical summer because they were selected and bred over millennia to withstand heat stress during tropical summer, so their FOT and SAD are weakly influenced by high day temperatures. However, the significant reduction in FED of these landraces, whose flowers open later in mid-day, seems to be an adaptive mechanism to avoid longer exposure to rising air temperature approaching day maximum temperature
Algorithm metadata vocabulary: A representational model and metadata vocabulary for describing and maintaining algorithms
No full textMetadata vocabularies are used in various domains of study. It provides an in-depth description of the resources. In this work, we develop algorithm metadata vocabulary (AMV), a vocabulary for capturing and storing the metadata about the algorithms (a procedure or a set of rules that is followed step-by-step to solve a problem, especially by a computer). The snag faced by the researchers in the current time is the failure of getting relevant results when searching for algorithms in any search engine. The designed vocabulary can be used by the algorithm repository developers, managers, and application developers. Besides, AMV is represented as a semantic model and produced OWL file, and it can be directly used by anyone interested to create and publish algorithm metadata as a knowledge graph, or to provide metadata service through the SPARQL endpoint. To design the vocabulary, we propose a well-defined methodology, which considers factual issues faced by the algorithm users and the practitioners. The evaluation shows promising results
An EEG-based neuro-recommendation system for improving consumer purchase experience
No full textThe aperture between the marketing domain and the electroencephalography (EEG)-based brain–computer interface (BCI) has been achieved with the inception of neuromarketing. This domain helps access the hidden information of the preferences and tastes of the consumers who intend to purchase. Research scholars have experimented with this emerging area in multiple aspects like designing pricing, promotions, predicting purchase-related activities, new product development, and so on. In this study, we have proposed an innovative use of neuromarketing to build a recommendation system. This recommendation system can potentially suggest suitable products to the consumer based on the past purchase behavior. This proposal carries huge potential in converting visitors to shoppers, increasing average order value, increasing the number of items per order, designing personalized promotions, and so on. The commonality of activated brain signals has been used to build this recommendation system. This neuromarketing-based recommendation system carries the advantage over the traditional recommendation system as this system suggests products based on the actual real-time state of the brain during the purchase. This system successfully initiated the starting point of building a neuromarketing-based recommendation system