arXiv subjects

Shweta Singh

Publications and source records attributed to Shweta Singh.

18 recordsLinked to original sources

arXiv2025

JobSphere: An AI-Powered Multilingual Career Copilot for Government Employment Platforms

Users of government employment websites commonly face engagement and accessibility challenges linked to navigational complexity, a dearth of language options, and a lack of personalized support. This paper introduces JobSphere, an AI-powered career assistant that is redefining the employment platform in Punjab called PGRKAM. JobSphere employs Retrieval-Augmented Generation (RAG) architecture, and it is multilingual, available in English, Hindi and Punjabi. JobSphere technique uses 4-bit quantization, allowing the platform to deploy on consumer-grade GPUs (i.e., NVIDIA RTX 3050 4GB), making the implementation 89% cheaper than that of cloud-based systems. Key innovations include voice-enabled interaction with the assistant, automated mock tests, resume parsing with skills recognition, and embed-based job recommendation that achieves a precision@10 score of 68%. An evaluation of JobSphere's implementation reveals 94% factual accuracy, a median response time of 1.8 seconds, and a System Usability Scale score of 78.5/100, a 50% improvement compared to the baseline PGRKAM platform context. In conclusion, JobSphere effectively fills significant accessibility gaps for Punjab/Hindi-speaking users in rural locations, while also affirming the users access to trusted job content provided by government agencies.

cs.AI↗

arXiv2025

Open-DeBias: Toward Mitigating Open-Set Bias in Language Models

Large Language Models (LLMs) have achieved remarkable success on question answering (QA) tasks, yet they often encode harmful biases that compromise fairness and trustworthiness. Most existing bias mitigation approaches are restricted to predefined categories, limiting their ability to address novel or context-specific emergent biases. To bridge this gap, we tackle the novel problem of open-set bias detection and mitigation in text-based QA. We introduce OpenBiasBench, a comprehensive benchmark designed to evaluate biases across a wide range of categories and subgroups, encompassing both known and previously unseen biases. Additionally, we propose Open-DeBias, a novel, data-efficient, and parameter-efficient debiasing method that leverages adapter modules to mitigate existing social and stereotypical biases while generalizing to unseen ones. Compared to the state-of-the-art BMBI method, Open-DeBias improves QA accuracy on BBQ dataset by nearly $48\%$ on ambiguous subsets and $6\%$ on disambiguated ones, using adapters fine-tuned on just a small fraction of the training data. Remarkably, the same adapters, in a zero-shot transfer to Korean BBQ, achieve $84\%$ accuracy, demonstrating robust language-agnostic generalization. Through extensive evaluation, we also validate the effectiveness of Open-DeBias across a broad range of NLP tasks, including StereoSet and CrowS-Pairs, highlighting its robustness, multilingual strength, and suitability for general-purpose, open-domain bias mitigation. The project page is available at: https://sites.google.com/view/open-debias25

cs.CL↗

arXiv2025

The power of dynamic causality in observer-based design for soft sensor applications

This paper introduces a novel framework for optimizing observer-based soft sensors through dynamic causality analysis. Traditional approaches to sensor selection often rely on linearized observability indices or statistical correlations that fail to capture the temporal evolution of complex systems. We address this gap by leveraging liquid-time constant (LTC) networks, continuous-time neural architectures with input-dependent time constants, to systematically identify and prune sensor inputs with minimal causal influence on state estimation. Our methodology implements an iterative workflow: training an LTC observer on candidate inputs, quantifying each input's causal impact through controlled perturbation analysis, removing inputs with negligible effect, and retraining until performance degradation occurs. We demonstrate this approach on three mechanistic testbeds representing distinct physical domains: a harmonically forced spring-mass-damper system, a nonlinear continuous stirred-tank reactor, and a predator-prey model following the structure of the Lotka-Volterra model, but with seasonal forcing and added complexity. Results show that our causality-guided pruning consistently identifies minimal sensor sets that align with underlying physics while improving prediction accuracy. The framework automatically distinguishes essential physical measurements from noise and determines when derived interaction terms provide complementary versus redundant information. Beyond computational efficiency, this approach enhances interpretability by grounding sensor selection decisions in dynamic causal relationships rather than static correlations, offering significant benefits for soft sensing applications across process engineering, ecological monitoring, and agricultural domains.

cs.AI↗

arXiv2025

Estimating the spatial economic and environmental impact of planned offshore wind energy in the USA using Environmentally Extended Multiregional Input-Output analysis

There is a projected increase in offshore wind energy generation in the United States over the next three decades, driven by legislative commitments and government funding. Like other renewable technologies, the construction of offshore wind farms has environmental impacts and spillover effects that must be assessed. Developing offshore wind as a reliable domestic energy source requires a multiregional analysis of economic and environmental effects of constructing projects along lakefronts and coastal regions. Although no commercial offshore wind farms currently operate in the United States, seven states have announced capacity commitments exceeding 28 gigawatts by 2035. This study evaluates the spatial economic and environmental impacts of planned projects by linking the National Renewable Energy Laboratory Offshore Renewables Balance-of-system Installation Tool (ORBIT) with a multiregional input-output model of the U.S. economy developed in the Virtual Industrial Ecology Lab. ORBIT provides capital investment requirements for installation, which are combined with the model to estimate economic spillover effects. Environmental impacts are assessed using a newly developed multiregional greenhouse gas emissions dataset for the U.S. to capture supply chain emissions of offshore wind construction. The five projects analyzed require 16.3 billion dollars in capital investment and generate 27.6 billion dollars in direct and indirect economic impacts across the country. Emissions results show that states active in energy generation are most affected, but impacts can be reduced by decarbonizing the grid. A carbon payback analysis indicates the projects offset construction-phase emissions in less than a year. The framework highlights which states experience the greatest spillover effects in terms of emissions and economic activity required to support offshore wind expansion.

econ.GN↗

arXiv2025

Cross talk between experimental data and simple validation of shell closure in pre-actinides

Two back-to-back experiments, 28Si + 178Hf and 28Si + 186W, were intentionally conducted to validate the role of shell closure in pre-actinides by studying neutron multiplicity in compound nucleus (CN) 206Rn and 214Ra. In the first experiment, Dubey et al. [Phys. Rev. C 112, L011602 (2025)], we established the influence of the neutron shell closure. In the present work, the CN 214Ra was deliberately selected to investigate the dependence of the total neutron multiplicity (Mtotal) on the proton number (Z), while keeping the neutron number constant at N = 126 in the pre-actinide region. The objective of the study is two-fold : (i) to examine the effect of proton shell closure when moving away from Z = 82, and (ii) to correlate the present results on proton shell closure with our previous finding on neutron shell closure. We have also used the previous reported data for N = 126 isotones 210Po, 212Rn, and 213Fr to establish the validation of shell closure. A systematic increase in Mtotal with increasing Z was observed from Z = 82 to Z = 88. Furthermore, comparison of the present results on proton shell closure with our earlier neutron shell closure observation, reveals that the cross-correlation between neutron and proton shell closure shows a systematic increase in Mtotal as one moves away from 208Pb, whether along isotonic or isotopic chains.

nucl-ex↗

arXiv2025

MSAD-Net: Multiscale and Spatial Attention-based Dense Network for Lung Cancer Classification

Lung cancer, a severe form of malignant tumor that originates in the tissues of the lungs, can be fatal if not detected in its early stages. It ranks among the top causes of cancer-related mortality worldwide. Detecting lung cancer manually using chest X-Ray image or Computational Tomography (CT) scans image poses significant challenges for radiologists. Hence, there is a need for automatic diagnosis system of lung cancers from radiology images. With the recent emergence of deep learning, particularly through Convolutional Neural Networks (CNNs), the automated detection of lung cancer has become a much simpler task. Nevertheless, numerous researchers have addressed that the performance of conventional CNNs may be hindered due to class imbalance issue, which is prevalent in medical images. In this research work, we have proposed a novel CNN architecture ``Multi-Scale Dense Network (MSD-Net)'' (trained-from-scratch). The novelties we bring in the proposed model are (I) We introduce novel dense modules in the 4th block and 5th block of the CNN model. We have leveraged 3 depthwise separable convolutional (DWSC) layers, and one 1x1 convolutional layer in each dense module, in order to reduce complexity of the model considerably. (II) Additionally, we have incorporated one skip connection from 3rd block to 5th block and one parallel branch connection from 4th block to Global Average Pooling (GAP) layer. We have utilized dilated convolutional layer (with dilation rate=2) in the last parallel branch in order to extract multi-scale features. Extensive experiments reveal that our proposed model has outperformed latest CNN model ConvNext-Tiny, recent trend Vision Transformer (ViT), Pooling-based ViT (PiT), and other existing models by significant margins.

cs.CV↗

arXiv2025

To examine the variation in dissipation near the shell closure using neutron multiplicity as a probe

The pre and post-scission neutron multiplicities have been determined for the fission of the compound nucleus (CN) 206Rn, induced by the reaction 28Si+178Hf within the excitation energy interval of 61.0-90.0 MeV. We intentionally formed CN 206 Rn, which is below the shell closure CN, to examine the variation in N/Z with total neutron multiplicity, as data for other CNs of 208,210,212,214,216Rn have already been published in the literature. We identified a new trend in the N/Z ratio, where the total neutron multiplicity initially decreases as we approach the shell closure of the compound nucleus and then starts to increase as we move away from the shell closure. Furthermore, we have observed that below the neutron shell closure, the dissipation in compound nuclei (CN) escalates with rising excitation energy, remains stable at the shell closure CN, and thereafter diminishes with increasing excitation energy above the shell closure CN.

nucl-ex↗

arXiv2025

IPO: Your Language Model is Secretly a Preference Classifier

Reinforcement learning from human feedback (RLHF) has emerged as the primary method for aligning large language models (LLMs) with human preferences. While it enables LLMs to achieve human-level alignment, it often incurs significant computational and financial costs due to its reliance on training external reward models or human-labeled preferences. In this work, we propose Implicit Preference Optimization (IPO), an alternative approach that leverages generative LLMs as preference classifiers, thereby reducing the dependence on external human feedback or reward models to obtain preferences. We conduct a comprehensive evaluation on the preference classification ability of LLMs using RewardBench, assessing models across different sizes, architectures, and training levels to validate our hypothesis. Furthermore, we investigate the self-improvement capabilities of LLMs by generating multiple responses for a given instruction and employing the model itself as a preference classifier for Direct Preference Optimization (DPO)-based training. Our findings demonstrate that models trained through IPO achieve performance comparable to those utilizing state-of-the-art reward models for obtaining preferences.

cs.CL↗

arXiv2024

Measurement of neutron induced reaction cross-section of tantalum with covariance analysis

The current study presents the cross-section measurement of $^{181}$Ta(n,$\gamma$)$^{182}$Ta reaction at 1.37 $\pm$ 0.13, 2.06 $\pm$ 0.14, 2.56 $\pm$ 0.15, and 3.05 $\pm$ 0.17 MeV neutron energies utilizing offline $\gamma$-ray spectroscopy. The neutrons were generated through the $^{7}$Li(p,n)$^{7}$Be reaction. The $^{115}$In(n,n'$\gamma$)$^{115m}$In reaction served as a monitor reaction. The covariance analysis was used to quantify the uncertainties in the measured cross-sections for the first time for the $^{181}$Ta(n,$\gamma$)$^{182}$Ta reaction. The present study provides detailed information on the propagation of uncertainty in the overall result. The required corrections for low energy background neutron and $\gamma$-ray coincidence summing effect have been made in the present measurement. The output is compared with the pre-existing cross-section data from the EXFOR database, evaluated data libraries and theoretical model predictions.

nucl-ex↗

arXiv2024

Resilience Dynamics in Coupled Natural-Industrial Systems: A Surrogate Modeling Approach for Assessing Climate Change Impacts on Industrial Ecosystems

Industrial ecosystems are coupled with natural systems through utilization of feedstocks and waste disposal. To ensure resilience in production of industrial systems under the threat of climate change scenarios, it is necessary to evaluate the impact of this coupling on productivity and waste generation. In this work, we present a novel methodology for modeling and assessing the resilience of coupled natural-industrial ecosystems under climate change scenarios. We develop a computationally efficient framework that integrates liquid time-constant (LTC) neural networks as surrogate models to capture complex, nonlinear dynamics of coupled agricultural and industrial systems. The approach is demonstrated through a case study of a soybean-based biodiesel production network in Champaign County, Illinois. LTC models are trained to capture dynamics of nodes and are then coupled and driven by statistically downscaled climate projections for RCP 4.5 and 8.5 scenarios from 2006-2096. The framework enables rapid simulation of system-wide material flow dynamics and exploration of cascading effects from climate-induced disruptions. Results reveal non-linear behaviors and potential tipping points in system resilience under different climate scenarios and farm sizes. The RCP 8.5 scenario led to earlier and more frequent production failures, increased reliance on imports for smaller farms, and complex patterns of waste accumulation and stock levels. The methodology provides valuable insights into system vulnerabilities and adaptive capacities, offering decision support for enhancing the resilience and sustainability of coupled natural-industrial ecosystems in the face of climate change. The framework's adaptability suggests potential applications across various industrial ecosystems and climate-sensitive sectors

eess.SY↗

arXiv2024

SKIPNet: Spatial Attention Skip Connections for Enhanced Brain Tumor Classification

Early detection of brain tumors through magnetic resonance imaging (MRI) is essential for timely treatment, yet access to diagnostic facilities remains limited in remote areas. Gliomas, the most common primary brain tumors, arise from the carcinogenesis of glial cells in the brain and spinal cord, with glioblastoma patients having a median survival time of less than 14 months. MRI serves as a non-invasive and effective method for tumor detection, but manual segmentation of brain MRI scans has traditionally been a labor-intensive task for neuroradiologists. Recent advancements in computer-aided design (CAD), machine learning (ML), and deep learning (DL) offer promising solutions for automating this process. This study proposes an automated deep learning model for brain tumor detection and classification using MRI data. The model, incorporating spatial attention, achieved 96.90% accuracy, enhancing the aggregation of contextual information for better pattern recognition. Experimental results demonstrate that the proposed approach outperforms baseline models, highlighting its robustness and potential for advancing automated MRI-based brain tumor analysis.

eess.IV↗

arXiv2024

Dynamic X-ray coherent diffraction analysis: bridging the timescales between imaging and photon correlation spectroscopy

The advent of diffraction limited sources and developments in detector technology opens up new possibilities for the study of materials in situ and operando. Coherent X-ray diffraction techniques such as coherent X-ray diffractive imaging (CXDI) and X-ray photon correlation spectroscopy (XPCS) are capable for this purpose and provide complimentary information, although due to signal-to-noise requirements, their simultaneous demonstration has been limited. Here, we demonstrate a strategy for the simultaneous use of CXDI and XPCS to study in situ the Brownian motion of colloidal gold nanoparticles of 200 nm diameter suspended in a glycerol-water mixture. We visualise the process of agglomeration, examine the spatiotemporal space accessible with the combination of techniques, and demonstrate CXDI with 22 ms temporal resolution.

cond-mat.mes-hall↗

arXiv2024

Benchmarking Object Detectors with COCO: A New Path Forward

The Common Objects in Context (COCO) dataset has been instrumental in benchmarking object detectors over the past decade. Like every dataset, COCO contains subtle errors and imperfections stemming from its annotation procedure. With the advent of high-performing models, we ask whether these errors of COCO are hindering its utility in reliably benchmarking further progress. In search for an answer, we inspect thousands of masks from COCO (2017 version) and uncover different types of errors such as imprecise mask boundaries, non-exhaustively annotated instances, and mislabeled masks. Due to the prevalence of COCO, we choose to correct these errors to maintain continuity with prior research. We develop COCO-ReM (Refined Masks), a cleaner set of annotations with visibly better mask quality than COCO-2017. We evaluate fifty object detectors and find that models that predict visually sharper masks score higher on COCO-ReM, affirming that they were being incorrectly penalized due to errors in COCO-2017. Moreover, our models trained using COCO-ReM converge faster and score higher than their larger variants trained using COCO-2017, highlighting the importance of data quality in improving object detectors. With these findings, we advocate using COCO-ReM for future object detection research. Our dataset is available at https://cocorem.xyz

cs.CV↗

arXiv2023

Revealing the impact of polystyrene-functionalization of Au octahedral nanocrystals of different sizes on formation and structure of mesocrystals

The self-assembly of anisotropic nanocrystals (stabilized by organic capping molecules) with pre-selected composition, size, and shape allows for the creation of nanostructured materials with unique structures and features. For such a material, the shape and packing of the individual nanoparticles play an important role. This work presents a synthesis procedure for ω-thiol-terminated polystyrene (PS-SH) functionalized gold nanooctahedra of variable size (edge length 37, 46, 58, and 72 nm). The impact of polymer chain length (Mw: 11k, 22k, 43k, and 66k g/mol) on the growth of colloidal crystals (e.g. mesocrystals) and their resulting crystal structure is investigated. Small-angle X-ray scattering (SAXS) and scanning transmission electron microscopy (STEM) methods provide a detailed structural examination of the self-assembled faceted mesocrystals based on octahedral gold nanoparticles of different size and surface functionalization. Three-dimensional angular X-ray cross-correlation analysis (AXCCA) enables high-precision determination of the superlattice structure and relative orientation of nanoparticles in mesocrystals. This approach allows us to perform non-destructive characterization of mesocrystalline materials and reveals their structure with resolution down to the nanometer scale.

cond-mat.mtrl-sci↗

arXiv2021

Reduced Order Dynamical Models For Complex Dynamics in Manufacturing and Natural Systems Using Machine Learning

Dynamical analysis of manufacturing and natural systems provides critical information about production of manufactured and natural resources respectively, thus playing an important role in assessing sustainability of these systems. However, current dynamic models for these systems exist as mechanistic models, simulation of which is computationally intensive and does not provide a simplified understanding of the mechanisms driving the overall dynamics. For such systems, lower-order models can prove useful to enable sustainability analysis through coupled dynamical analysis. There have been few attempts at finding low-order models of manufacturing and natural systems, with existing work focused on model development of individual mechanism level. This work seeks to fill this current gap in the literature of developing simplified dynamical models for these systems by developing reduced-order models using a machine learning (ML) approach. The approach is demonstrated on an entire soybean-oil to soybean-diesel process plant and a lake system. We use a grey-box ML method with a standard nonlinear optimization approach to identify relevant models of governing dynamics as ODEs using the data simulated from mechanistic models. Results show that the method identifies a high accuracy linear ODE models for the process plant, reflective of underlying linear stoichiometric mechanisms and mass balance driving the dynamics. For the natural systems, we modify the ML approach to include the effect of past dynamics, which gives non-linear ODE. While the modified approach provides a better match to dynamics of stream flow, it falls short of completely recreating the dynamics. We conclude that the proposed ML approach work well for systems where dynamics is smooth, such as in manufacturing plant whereas does not work perfectly well in case of chaotic dynamics such as water stream flow.

eess.SY↗

arXiv2019

Can Machine Learning Identify Governing Laws For Dynamics in Complex Engineered Systems ? : A Study in Chemical Engineering

Machine learning recently has been used to identify the governing equations for dynamics in physical systems. The promising results from applications on systems such as fluid dynamics and chemical kinetics inspire further investigation of these methods on complex engineered systems. Dynamics of these systems play a crucial role in design and operations. Hence, it would be advantageous to learn about the mechanisms that may be driving the complex dynamics of systems. In this work, our research question was aimed at addressing this open question about applicability and usefulness of novel machine learning approach in identifying the governing dynamical equations for engineered systems. We focused on distillation column which is an ubiquitous unit operation in chemical engineering and demonstrates complex dynamics i.e. it's dynamics is a combination of heuristics and fundamental physical laws. We tested the method of Sparse Identification of Non-Linear Dynamics (SINDy) because of it's ability to produce white-box models with terms that can be used for physical interpretation of dynamics. Time series data for dynamics was generated from simulation of distillation column using ASPEN Dynamics. One promising result was reduction of number of equations for dynamic simulation from 1000s in ASPEN to only 13 - one for each state variable. Prediction accuracy was high on the test data from system within the perturbation range, however outside perturbation range equations did not perform well. In terms of physical law extraction, some terms were interpretable as related to Fick's law of diffusion (with concentration terms) and Henry's law (with ratio of concentration and pressure terms). While some terms were interpretable, we conclude that more research is needed on combining engineering systems with machine learning approach to improve understanding of governing laws for unknown dynamics.

eess.SY↗

arXiv2012

Adjacency Matrix Based Energy Efficient Scheduling using S-MAC Protocol in Wireless Sensor Networks

Communication is the main motive in any Networks whether it is Wireless Sensor Network, Ad-Hoc networks, Mobile Networks, Wired Networks, Local Area Network, Metropolitan Area Network, Wireless Area Network etc, hence it must be energy efficient. The main parameters for energy efficient communication are maximizing network lifetime, saving energy at the different nodes, sending the packets in minimum time delay, higher throughput etc. This paper focuses mainly on the energy efficient communication with the help of Adjacency Matrix in the Wireless Sensor Networks. The energy efficient scheduling can be done by putting the idle node in to sleep node so energy at the idle node can be saved. The proposed model in this paper first forms the adjacency matrix and broadcasts the information about the total number of existing nodes with depths to the other nodes in the same cluster from controller node. When every node receives the node information about the other nodes for same cluster they communicate based on the shortest depths and schedules the idle node in to sleep mode for a specific time threshold so energy at the idle nodes can be saved.

cs.NI↗

arXiv2011

Incorporating Agile with MDA Case Study: Online Polling System

Nowadays agile software development is used in greater extend but for small organizations only, whereas MDA is suitable for large organizations but yet not standardized. In this paper the pros and cons of Model Driven Architecture (MDA) and Extreme programming have been discussed. As both of them have some limitations and cannot be used in both large scale and small scale organizations a new architecture has been proposed. In this model it is tried to opt the advantages and important values to overcome the limitations of both the software development procedures. In support to the proposed architecture the implementation of it on Online Polling System has been discussed and all the phases of software development have been explained.

cs.SE↗