ISI Digital Commons (Indian Statistical Institute )
Not a member yet
7571 research outputs found
Sort by
A complex network analysis approach to compare the performance of batsmen across different formats
No full textBatsmen are accorded a certain precedence for better batting ability over their peers. The batsmen in cricket are assessed mostly based on their batting average. However, comparing players by batting average over different timelines does not yield the appropriate results, as the batting productivity of a particular player varies in unique ways across the different formats of the game. Using batting averages for comparison does not include factors such as the speed of scoring runs and the frequency of milestones achieved. The objective of this study is to present an effective knowledge-based mechanism for judging and comparing the batting strength of players in different formats of cricket. This methodology uses a complex network consisting of effective features that are subsequently integrated to formulate a Batting Precedence Score, which is further incorporated into an efficient Batting Precedence Score algorithm. In addition, we created a structured World Wide Batsman Dataset (WWBD) for our analysis based on the ESPN Cricinfo dataset. The results of extensive experiments demonstrate that the proposed method provides promising insights. The batting precedence of the proposed method is further compared with those of existing methods, and the proposed method displays better results
Risk assessment of trace elements in vegetables grown in river Yamuna floodplain in Delhi
No full textUrban agriculture is common in fertile river floodplains of many developing countries. However, there is a risk of contamination in highly polluted regions. This study quantifies health risks associated with the consumption of vegetables grown in the floodplain of the urban river ‘Yamuna’ in the highly polluted yet data-scarce megacity Delhi, India. Six trace elements are analyzed in five kinds of vegetable samples. Soil samples from the cultivation area are also analyzed for elemental contamination. Ni, Mn, and Co are observed to be higher in leafy vegetables than others. Fruit and inflorescence vegetables are found to have higher concentrations of Cr, Pb, and Zn as compared to root vegetables. Transfer Factor indicates that Cr and Co have the highest and least mobility, respectively. Vegetable Pollution Index indicates that contamination levels follow as Cr \u3e Ni \u3e Pb \u3e Zn. Higher Metal Pollution Index of leafy and inflorescence vegetables than root and fruit vegetables indicate that atmospheric deposition is the predominant source. Principal Component Analysis indicates that Pb and Cr have similar sources and patterns in accumulation. Among the analyzed vegetables, radish may pose a non-carcinogenic risk to the age group of 1–5 year. Carcinogenic risk is found to be potentially high due to Ni and Cr accumulation. Consumption of leafy vegetables was found to have relatively less risk than other vegetables due to lower Cr accumulation. Remediation of Cr and Ni in floodplain soil and regular monitoring of elemental contamination is a priority
Securing Reusable IP Cores Using Voice Biometric Based Watermark
No full textReusable third-party intellectual property (3PIP) cores within the supply chain are vulnerable to hardware threats such as IP piracy and false claim of ownership. Securing the reusable IP cores is vital to protect the original vendor from a substantial revenue loss and his/her brand value. This paper presents a novel hardware IP core watermarking methodology based on voice biometric signature to enable detective control against IP piracy and resolve IP ownership claim. To the best of our knowledge, this is the first voice biometric-based hardware IP protection technique. This paper proposes a novel methodology for generating a unique voice signature template using distinct voice features, viz. jitter and shimmer, along with pitch and intensity values at different timestamps. We present a high-level synthesis (HLS) design methodology of embedding a voice signature digital template during the register allocation phase to generate secured IP cores. Results and analysis imply that the proposed approach can significantly improve security in terms of stronger authorship proof and higher tamper tolerance compared to the existing IP watermarking approaches. Additionally, we also analyze the uniqueness of a voice signature and its security against forgery attack. We achieve higher security at negligible design cost overhead
SemiDocSeg: harnessing semi-supervised learning for document layout analysis
No full textDocument Layout Analysis (DLA) is the process of automatically identifying and categorizing the structural components (e.g. Text, Figure, Table, etc.) within a document to extract meaningful content and establish the page’s layout structure. It is a crucial stage in document parsing, contributing to their comprehension. However, traditional DLA approaches often demand a significant volume of labeled training data, and the labor-intensive task of generating high-quality annotated training data poses a substantial challenge. In order to address this challenge, we proposed a semi-supervised setting that aims to perform learning on limited annotated categories by eliminating exhaustive and expensive mask annotations. The proposed setting is expected to be generalizable to novel categories as it learns the underlying positional information through a support set and class information through Co-Occurrence that can be generalized from annotated categories to novel categories. Here, we first extract features from the input image and support set with a shared multi-scale feature acquisition backbone. Then, the extracted feature representation is fed to the transformer encoder as a query. Later on, we utilize a semantic embedding network before the decoder to capture the underlying semantic relationships and similarities between different instances, enabling the model to make accurate predictions or classifications with only a limited amount of labeled data. Extensive experimentation on competitive benchmarks like PRIMA, DocLayNet, and Historical Japanese (HJ) demonstrate that this generalized setup obtains significant performance compared to the conventional supervised approach
Some further results on random OBIC rules
No full textWe study the structure of probabilistic voting rules that are ordinal Bayesian incentive compatible (OBIC) with respect to independently distributed prior beliefs that can be considered generic (Majumdar and Sen (2004)). We first identify a class of priors, such that for each prior in that class there exists a probabilistic voting rule that puts a positive probability weight on “compromise” candidates. The class of priors include generic priors. Next, we consider a class of randomized voting rules that have a “finite range”. For this class of rules, we identify an appropriate generic condition on priors such that, any rule in this class is OBIC with respect to a prior satisfying the generic condition if and only if the rule is a random dictatorship
Survey of C.R. Rao’s Orthogonal Arrays, Balanced Arrays, and Their Applications
No full textThis comprehensive review article on orthogonal arrays (OAs), balanced arrays (BAs) and their practical applications serves as a tribute to the life and ground breaking contributions of the legendary statistician, C.R. Rao (1920-2023). It highlights his profound influence on the field of statistical sciences and explores the significant contributions he made to the realms of OAs and BAs. His work in these areas has left an indelible impact on the domains of experimental design, combinatorial mathematics, and statistical analysis. In this article, we delve into some noteworthy applications of OAs and BAs
Unraveling the unknown: Adaptive spatial planning to enhance climate resilience for the endangered Swamp Grass-babbler (Laticilla cinerascens) with habitat connectivity and complexity approach
No full textThe endangered and poorly known Swamp Grass-babbler, Laticilla cinerascens (Passeriformes: Pellorneidae), confronts critical threats and vulnerability due to its specific habitat requirements and restricted populations in the northeastern region of the Indian Subcontinent. This study investigates the distribution of the species, habitat quality, geometry and shape complexity of connectivity among the protected areas (PAs), and responses to climate change in Northeast India under different climate change pathways by utilizing ensemble distribution models, and ecological metrics. From the total distribution extent (1,42,000 km2), approximately 9366 km2 (6.59 %) is identified as the suitable habitat for this threatened species. Historically centered around Dibru Saikhowa National Park (DSNP), the species faced a drastic decline due to anthropogenic activities and alteration in land use and lover cover. The study also reveals a significant decline in suitable habitat for L. cinerascens in future climate scenarios, with alarming reductions under SSP126 (\u3e10 % in the timeframe 2041–2060 and \u3e 30 % from 2061 to 2080), SSP245 (\u3e90 % in both time periods), and SSP585 (\u3e90 % in both timeframes) from the present scenario. At present, DSNP has the most suitable habitat within the distribution range but is projected to decline (\u3e90 %) under more severe climate change scenarios, as observed in other PAs. Landscape fragmentation analysis indicates a shift in habitat geometry, highlighting the intricate impact of climate change. It predicts a substantial 343 % increase (in the SSP126) in small habitat patches in the future. Connectivity analysis among PAs shows a significant shift, with a decline exceeding 20 %. The analysis of shape complexity and connectivity geometry reveals a significant increase of over 220 % in the fragmentation of connectivity among PAs between 2061 and 2080 under the SSP585 climate change scenario compared to the present conditions. The study underscores the urgent need for conservation actions, emphasizing the complex interplay of climate change, habitat suitability, and fragmentation. Prioritizing PAs with suitable habitats and assessing their connectivity is crucial. Adaptive management strategies are essential to address ongoing environmental changes and safeguard biodiversity. Future research in critical areas is needed to establish long-term monitoring programs to lead/extend effective conservation strategies
A New Unsupervised Approach for Text Localization in Shaky and Non-shaky Scene Video
No full textText Detection in shaky and non-shaky videos is challenging due to poor video quality and the presence of static and dynamic obstacles. Video captured by a shaky camera due to wind is considered shaky video, while video captured by a fixed camera is considered as non-shaky video. Most state-of-the-art methods achieve the best results when exploring the concept of deep learning. The present study proposes an unsupervised approach for text spotting in shaky and non-shaky videos. In the first stage, our method selects keyframes from the input video by estimating the similarity between the temporal frames, which we named activation frames. For each activation frame, the proposed method extracts statistical features such as orientation, spectral, edge density and intensity features that represent text information. The extracted features are fed to a K-means clustering method to obtain the text clusters, which results in text regions in the activation frames. For each region, the proposed method uses optical flow to extract spatial consistency, motion consistency and depth map consistency for localizing text using temporal voting-non-maximum suppression. Experiments are conducted on our shaky and non-shaky dataset, and the benchmark dataset of ICDAR 2015. For the experiments it can be seen that the proposed method is superior to existing methods
Digital Twin Technology for River Basin Management: A Framework for Proactive Flood Mitigation and Water Resource Optimization
No full textThe increasing frequency of extreme weather events due to climate change is leading to more incidents of river flooding and water clogging. Digital twin technology provides a promising solution for modelling and mitigating these issues by creating a virtual replica of the physical river system. This paper explores the development of a digital twin model specifically for monitoring and predicting water clogging and flow patterns in rivers. From acquiring real-time sensor data on river levels, precipitation, and environmental conditions to processing this multidimensional data, the digital twin model integrates physics-based hydraulic simulations with data-driven machine learning models. The 3D virtual environment precisely mirrors the river\u27s geometry, terrain, infrastructure like bridges and dams, and dynamic variables like water velocity and depth. The digital twin continuously maps the state of the physical river by assimilating live sensor feeds. Machine learning models calibrated on historical data help forecast river flow rates, water depths, potential clogging areas, and flood risks. Physics simulations incorporate these predictions along with the river\u27s characteristics to model the flow dynamics accurately. This synergy enables proactive monitoring, early warning systems, testing mitigation strategies, and optimizing reservoir operations. The proposed digital twin model for river clogging and flow lays the foundation for digital river basin engineering - an integrated pipeline spanning data acquisition, modelling, simulation, decision support, and control actions. The results can guide water resource management, urban planning near rivers, and disaster preparedness, ultimately enhancing resilience against flooding events
DOST—Domain Obedient Self-supervision for Trustworthy Multi Label Classification with Noisy Labels
No full textIncorporating the vast expertise of clinical practitioners into deep learning systems can massively improve trustworthiness and performance of these systems. Deep learning systems rely on enormous amounts of data, often accompanied by annotation errors, and do not natively abide by well-known medical principles. In diagnostic scenarios, lack of adherence to domain constraints make systems unreliable, and this problem is only exacerbated by annotation errors. This area has been relatively unexplored in the context of “multi-label classification” (MLC) tasks which feature more complex noise. This paper studies the effect of label noise on domain rule violation incidents, and incorporates clinical rules into our learning algorithm to improve trustworthiness. We propose the DOST paradigm and experimentally show that our approach not only makes deep learning models more aligned to domain rules, but also improves learning performance in key metrics and minimizes the effect of annotation noise. This novel approach uses domain guidance to detect offending annotations and deter rule-violating predictions in a self-supervised manner, thus making it more “data efficient” and domain compliant