11563 research outputs found
Sort by
COMI-LINGUA: expert annotated large-scale dataset for multitask NLP in Hindi-English code-mixing
No full textThe rapid growth of digital communication has driven the widespread use of code-mixing, particularly Hindi-English, in multilingual communities. Existing datasets often focus on romanized text, have limited scope, or rely on synthetic data, which fails to capture realworld language nuances. Human annotations are crucial for assessing the naturalness and acceptability of code-mixed text. To address these challenges, We introduce COMI-LINGUA, the largest manually annotated dataset for code-mixed text, comprising 100,970 instances evaluated by three expert annotators in both Devanagari and Roman scripts. The dataset supports five fundamental NLP tasks: Language Identification, Matrix Language Identification, Part-of-Speech Tagging, Named Entity Recognition, and Translation. We evaluate LLMs on these tasks using COMILINGUA, revealing limitations in current multilingual modeling strategies and emphasizing the need for improved code-mixed text processing capabilities. COMI-LINGUA is publically availabe at: this https://huggingface.co/datasets/LingoIITGN/COMI-LINGU
Nanosheet FET based CMOS technology for future electronics: challenges in sheet thickness scaling and solutions
No full textMulti-Day Extreme Precipitation Caused Major Floods in India During Summer Monsoon of 2024
No full textThe risk of extreme precipitation and flooding has increased in India due to climate change. During the 2024 summer monsoon season, three major extreme precipitation events occurred across the western, southern, and northern states of India, leading to widespread flooding in these regions. We examine the causes and impacts of extreme precipitation and flood events using a combination of observational data, reanalysis data sets, and hydrological models. In all the three regions, extreme rainfall occurred immediately after multiday continuous precipitation, resulting in catastrophic flooding. The 3-day extreme precipitation that caused flooding in the three regions had return periods of more than 75 years, 100 years, and 200 years, respectively. The primary moisture sources for the Gujarat floods were the Arabian Sea and the Indian Ocean, while the floods in Andhra Pradesh and Telangana were mainly sourced by the Bay of Bengal. For the floods in northern India, the dominant moisture sources were recycled land moisture and moisture transport from the Bay of Bengal. These moisture inflows, coupled with favorable atmospheric conditions, resulted in multiday extreme precipitation in the three regions. Saturated soil moisture conditions before the extreme precipitation contributed to high runoff, triggering extensive flooding in all the three regions. Our findings highlight the growing challenge of managing such extreme events as their frequency and intensity are projected to increase under a warming climate
Microbial enzyme in food biotechnology
No full textEnzymes are used in the improvement of processed food products. The first successful commercial use of enzyme in food processing was in the cheese-making process. Now enzymes are used in brewing, meat tenderization, baking, and protein hydrolysis, etc. Enzymes enhance nutritional value and flavor of ingredients and food products. Enzymes that are involved in food processing have an important role-to change the conventional, harmful, and chemical-based processes and replace them with eco-friendly processes, enhance the biodegradability of products, and lower energy consumption levels. It may be beneficial to unlock a new path for enzyme applications, which could be an advantage for the food-processing industry. Our hope is to change the moral value of community regarding recombinant DNA technology and enzyme (protein) engineering techniques, to raise the enzyme applications in a more advanced way in the food industry. � 2021 Elsevier B.V., All rights reserved
Sensitivity and query complexity under uncertainty
No full textIn this paper, we study the query complexity of Boolean functions in the presence of uncertainty, motivated by parallel computation with an unlimited number of processors where inputs are allowed to be unknown. We allow each query to produce three results: zero, one, or unknown. The output could also be: zero, one, or unknown, with the constraint that we should output ''unknown'' only when we cannot determine the answer from the revealed input bits. Such an extension of a Boolean function is called its hazard-free extension.
- We prove an analogue of Huang's celebrated sensitivity theorem [Annals of Mathematics, 2019] in our model of query complexity with uncertainty.
- We show that the deterministic query complexity of the hazard-free extension of a Boolean function is at most quadratic in its randomized query complexity and quartic in its quantum query complexity, improving upon the best-known bounds in the Boolean world.
- We exhibit an exponential gap between the smallest depth (size) of decision trees computing a Boolean function, and those computing its hazard-free extension.
- We present general methods to convert decision trees for Boolean functions to those for their hazard-free counterparts, and show optimality of this construction. We also parameterize this result by the maximum number of unknown values in the input.
- We show lower bounds on size complexity of decision trees for hazard-free extensions of Boolean functions in terms of the number of prime implicants and prime implicates of the underlying Boolean function
Design Space and Variability Analysis of SOI MOSFET for Ultralow-Power Band-to-Band Tunneling Neurons
No full textLarge spiking neural networks (SNNs) require ultralow power and low variability hardware for neuromorphic computing applications. Recently, a band-to-band tunneling (BTBT)-based integrator was proposed, enabling the sub-kHz operation of neurons with area and energy efficiency. For an ultralow-power implementation of such neurons, a very low BTBT current is needed, so minimizing current without degrading neuronal properties is essential. Low variability is needed in the ultralow current integrator to avoid network performance degradation in a large BTBT neuron-based SNN. This work addresses device optimization to achieve low BTBT current. We conducted design space and variability analysis in technology computer-aided design (TCAD), utilizing a well-calibrated TCAD deck with experimental data from GlobalFoundries (GFs) 32 nm partially depleted silicon-on-insulator (PD-SOI) MOSFET. First, we discuss the physics-based explanation of the tunneling mechanism. Second, we explore the impact of device design parameters on SOI MOSFET performance, highlighting parameter sensitivities to tunneling current. With device parameters' optimization, we demonstrate a ∼20× reduction in BTBT current compared to the experimental data. Finally, a variability analysis that includes the effects of random dopant fluctuations (RDFs), oxide thickness variation (OTV), and channel-oxide interface traps (DIT) in the BTBT, subthreshold (SS), and ON regimes of operation is shown. The BTBT regime shows the highest sensitivity to OTV, with variability increasing by up to 25× compared to the SS regime. In contrast, RDF and DIT variability resulted in a 1.25× to ∼10× lower coefficient of variation (CV) in the BTBT regime than in the SS regime, indicating better resilience to these sources of variability. The study provides essential design guidelines to enable energy-efficient neuromorphic computing, achieving biologically plausible sub-kHz spiking frequencies
Water Scarcity and Land Degradation Nexus in the Anthropocene: Reformations for Advanced Water Management as Per the Sustainable Development Goals
No full textImpact of surface patterning on oxygen vacancy formation and subsequent photoelectrochemical performance of TiO2 nanostructures
No full textThis research demonstrates unique approach for fabricating patterned titanium (Ti) substrates through micro-milling and using them to grow dual-topographic TiO2 Micro-Nanofibers by anodization. X-ray diffraction and Raman spectroscopy studies revealed that the TiO2 lattice on micro-milled substrate exhibited evident lattice expansion compared to TiO2 grown on the un-patterned Ti substrate. Photoemission studies reveal the oxygen vacancies, and Ti3+ states as key factors contributing to this expansion. The resulting Micro-Nanofibrous structures have higher surface area, oxygen vacancies, and enhanced light-harvesting capabilities, which are responsible for higher photocurrent and durability than TiO2 nanofibers grown on un-patterned Ti substrate. Under 1 sun illumination at 1.23 VRHE in 1 M KOH, these structures produced approximately 28 % higher photocurrent than their un-patterned counterparts. The micro-milling technique provides a cost-effective, rapid prototyping solution for generating stable nanostructures with microscale features offering advantages over traditional 3D printing and laser ablation techniques. This method shows great potential for photoelectrochemical water splitting
