arXiv subjects

Sumit Kumar

Publications and source records attributed to Sumit Kumar.

At least 37 records · Page 2Linked to original sources

arXiv2026

User-Centric Comparison of 5G NTN and DVB-S2/RCS2 Using OpenAirInterface and OpenSAND

The integration of satellite networks into next-generation mobile communication systems has gained considerable momentum with the advent of 5G Non-Terrestrial Networks (5G-NTN). Since established technologies like DVB-S2/RCS2 are already widely used for satellite broadband, a detailed comparison with emerging 5G NTN solutions is necessary to understand their relative merits and guide deployment decisions. This paper presents a user-centric, end-to-end evaluation of these technologies under realistic traffic conditions, showing how differences in architecture and protocols impact application-layer performance. Utilizing the 6G Sandbox platform, we employ OpenAirInterface to emulate 5G NTN and OpenSAND for DVB-S2/RCS2, replicating transparent payload GEO satellite scenarios under uniform downlink conditions. A range of real-world applications, such as web browsing, file downloads, and video streaming, are tested across both systems and systematically analyzed. While the emulation lacks real-time capability, it reveals key strengths and limitations of each approach, helping identify suitable deployment scenarios for 5G NTN and DVB-S2/RCS2.

cs.NI↗

arXiv2025

HEL1OS -- A Hard X-ray Spectrometer on Board Aditya-L1

HEL1OS (High Energy L1 Orbiting X-ray Spectrometer) is one of the remote sensing payloads on board Aditya-L1 mission designed to continuously monitor and measure the time-resolved spectra of solar flares between 8 keV and 150 keV. This broad energy range has been covered by using compound semiconductor detectors: cadmium telluride (CdTe: 8 - 70 keV) and cadmium zinc telluride (CZT: 20 - 150 keV) with geometric areas of 0.5 cm$^2$ and 32 cm$^2$, respectively. A stainless steel collimator provides a field-of-view of 6$^\circ$ $\times$ 6$^\circ$ optimized to limit the off-axis response while keeping the design within the instrument mass constraints. The in-house designed low-noise digital pulse processing-based front-end electronics has achieved a spectral resolution of $\approx$ 1 keV at 14 keV (CdTe) and $\approx$ 7 keV at 60 keV (CZT). The instrument is also equipped with processing and power electronics to process the signal, drive the electronics, bias the detectors with required low and high voltages for optimal performance of the overall system. In this article, we present design aspects of the instrument, results from the pre-launch ground-based tests, and the in-orbit operations, which have indicated optimal performance in line with that expected.

astro-ph.SR↗

arXiv2025

Malaria detection using Deep Convolution Neural Network

The latest WHO report showed that the number of malaria cases climbed to 219 million last year, two million higher than last year. The global efforts to fight malaria have hit a plateau and the most significant underlying reason is international funding has declined. Malaria, which is spread to people through the bites of infected female mosquitoes, occurs in 91 countries but about 90% of the cases and deaths are in sub-Saharan Africa. The disease killed 4,35,000 people last year, the majority of them children under five in Africa. AI-backed technology has revolutionized malaria detection in some regions of Africa and the future impact of such work can be revolutionary. The malaria Cell Image Data-set is taken from the official NIH Website NIH data. The aim of the collection of the dataset was to reduce the burden for microscopists in resource-constrained regions and improve diagnostic accuracy using an AI-based algorithm to detect and segment the red blood cells. The goal of this work is to show that the state of the art accuracy can be obtained even by using 2 layer convolution network and show a new baseline in Malaria detection efforts using AI.

eess.IV↗

arXiv2025

Data efficient surrogate modeling for engineering design: Ensemble-free batch mode deep active learning for regression

High fidelity design evaluation processes such as Computational Fluid Dynamics and Finite Element Analysis are often replaced with data driven surrogates to reduce computational cost in engineering design optimization. However, building accurate surrogate models still requires a large number of expensive simulations. To address this challenge, we introduce epsilon HQS, a scalable active learning strategy that leverages a student teacher framework to train deep neural networks efficiently. Unlike Bayesian AL methods, which are computationally demanding with DNNs, epsilon HQS selectively queries informative samples to reduce labeling cost. Applied to CFD, FEA, and propeller design tasks, our method achieves higher accuracy under fixed labeling cost budgets.

cs.LG↗

arXiv2025

Accounting for the Known Unknowns: A Parametric Framework to Incorporate Systematic Waveform Errors in Gravitational-Wave Parameter Estimation

The PE for GW merger events relies on a waveform model calibrated using numerical simulations. Within the Bayesian framework, this waveform model represents the GW signal produced during the merger and is crucial for estimating the likelihood function. However, these waveform models may possess systematic errors that can differ across the parameter space. Addressing these errors in the current data analysis pipeline is an active area of research. We introduce parametrizations for the uncertainties in the amplitude and phase of the reference waveform model. When the error budget in the amplitude and phase of the waveform model, as a function of frequency, is known, it can be used as a prior distribution in the Bayesian framework. We also show that conservative priors can be used to quantify uncertainties in waveform modeling without any knowledge of waveform uncertainty error budgets. Through zero-noise injections and PE recoveries, we demonstrate that even 1%-2% of errors in relative phase to the actual waveform model, for a GW150914-like signal and advanced LIGO detector sensitivity, can introduce biases in the recovered parameters. These biases can be corrected when we account for waveform uncertainties within the PE framework. By analyzing a series of simulated signals from mergers with precessing orbits and recovering them using a non-spinning waveform model, we demonstrate that we can reduce the ratio of systematic errors to statistical errors. This approach allows us to address scenarios where specific physical effects are missing in waveform modeling. The code that implements our parametrization for performing PE is available as a Python package "pycbc_wferrors_plugin", compatible with the PyCBC open source GW analysis library.

gr-qc↗

arXiv2025

Portfolio construction using a sampling-based variational quantum scheme

The efficient and effective construction of portfolios that adhere to real-world constraints is a challenging optimization task in finance. We investigate a concrete representation of the problem with a focus on design proposals of an Exchange Traded Fund. We evaluate the sampling-based CVaR Variational Quantum Algorithm (VQA), combined with a local-search post-processing, for solving problem instances that beyond a certain size become classically hard. We also propose a problem formulation that is suited for sampling-based VQA. Our utility-scale experiments on IBM Heron processors involve 109 qubits and up to 4200 gates, achieving a relative solution error of 0.49%. Results indicate that a combined quantum-classical workflow achieves better accuracy compared to purely classical local search, and that hard-to-simulate quantum circuits may lead to better convergence than simpler circuits. Our work paves the path to further explore portfolio construction with quantum computers.

quant-ph↗

arXiv2025

Evaluating Open-Source Large Language Models for Technical Telecom Question Answering

Large Language Models (LLMs) have shown remarkable capabilities across various fields. However, their performance in technical domains such as telecommunications remains underexplored. This paper evaluates two open-source LLMs, Gemma 3 27B and DeepSeek R1 32B, on factual and reasoning-based questions derived from advanced wireless communications material. We construct a benchmark of 105 question-answer pairs and assess performance using lexical metrics, semantic similarity, and LLM-as-a-judge scoring. We also analyze consistency, judgment reliability, and hallucination through source attribution and score variance. Results show that Gemma excels in semantic fidelity and LLM-rated correctness, while DeepSeek demonstrates slightly higher lexical consistency. Additional findings highlight current limitations in telecom applications and the need for domain-adapted models to support trustworthy Artificial Intelligence (AI) assistants in engineering.

cs.NI↗

arXiv2025

LABELING COPILOT: A Deep Research Agent for Automated Data Curation in Computer Vision

Curating high-quality, domain-specific datasets is a major bottleneck for deploying robust vision systems, requiring complex trade-offs between data quality, diversity, and cost when researching vast, unlabeled data lakes. We introduce Labeling Copilot, the first data curation deep research agent for computer vision. A central orchestrator agent, powered by a large multimodal language model, uses multi-step reasoning to execute specialized tools across three core capabilities: (1) Calibrated Discovery sources relevant, in-distribution data from large repositories; (2) Controllable Synthesis generates novel data for rare scenarios with robust filtering; and (3) Consensus Annotation produces accurate labels by orchestrating multiple foundation models via a novel consensus mechanism incorporating non-maximum suppression and voting. Our large-scale validation proves the effectiveness of Labeling Copilot's components. The Consensus Annotation module excels at object discovery: on the dense COCO dataset, it averages 14.2 candidate proposals per image-nearly double the 7.4 ground-truth objects-achieving a final annotation mAP of 37.1%. On the web-scale Open Images dataset, it navigated extreme class imbalance to discover 903 new bounding box categories, expanding its capability to over 1500 total. Concurrently, our Calibrated Discovery tool, tested at a 10-million sample scale, features an active learning strategy that is up to 40x more computationally efficient than alternatives with equivalent sample efficiency. These experiments validate that an agentic workflow with optimized, scalable tools provides a robust foundation for curating industrial-scale datasets.

cs.CV↗

arXiv2025

SEMMA: A Semantic Aware Knowledge Graph Foundation Model

Knowledge Graph Foundation Models (KGFMs) have shown promise in enabling zero-shot reasoning over unseen graphs by learning transferable patterns. However, most existing KGFMs rely solely on graph structure, overlooking the rich semantic signals encoded in textual attributes. We introduce SEMMA, a dual-module KGFM that systematically integrates transferable textual semantics alongside structure. SEMMA leverages Large Language Models (LLMs) to enrich relation identifiers, generating semantic embeddings that subsequently form a textual relation graph, which is fused with the structural component. Across 54 diverse KGs, SEMMA outperforms purely structural baselines like ULTRA in fully inductive link prediction. Crucially, we show that in more challenging generalization settings, where the test-time relation vocabulary is entirely unseen, structural methods collapse while SEMMA is 2x more effective. Our findings demonstrate that textual semantics are critical for generalization in settings where structure alone fails, highlighting the need for foundation models that unify structural and linguistic signals in knowledge reasoning.

cs.CL↗

arXiv2025

Density Functional Theory Analysis of Na3AgO: Assessing its Viability as a Sustainable Material for Solar Energy Applications

This study mainly emphasis the fascinating features of inverse perovskites Na3AgO using density functional theory (DFT). Inverse perovskite (IP) Na3AgO structural features have been examined, and the space group and cubic structure of Pm-3m (221) have been confirmed. The experimental formulation and thermal stability of IP have been confirmed by the formation energy. Phonon dispersion curves were used to assess dynamic stability. The dynamic stability of the examined IP and the bonding strength against cubic structure deformation are confirmed by the lack of negative frequencies. The energy gap or the characteristics of semiconducting behaviour have been predicted by the electronic properties of Na3AgO with a band gap of 1.273 eV. In order to confirmthe viability of solar cells, the light-dependent properties have also been identified. Born stability criteria are also used to verify the mechanical stability, and additional elastic characteristics are identified in order to forecast the anisotropy, ductility, strength, and hardness. These anti-perovskites, which possess intriguing characteristics, have the potential to be effective materials for photovoltaic applications, as indicated by the analysed findings.

cond-mat.mtrl-sci↗

arXiv2025

Exploring PdCrAs Half-Heusler Alloy for Sustainable Energy Solutions: An Ab-initio Study

This work presents a comprehensive investigation of the HH alloy PdCrAs using first - principles methods, highlighting its potential applications in various fields, including spintronics, thermoelectrics, and optoelectronics. We employed density functional theory (DFT) within the full potential linearized augmented plane wave (FLAPW) framework. Structural optimizations indicate that the alloy stabilizes in the ferromagnetic phase. Both mechanical and dynamical stability have been confirmed through analyses of elastic constants and phonon dispersion. Our calculations of the electronic band structure and density of states (DOS) reveal that PdCrAs exhibits half-metallic behavior, with a spin-polarized band gap of 0.670 eV in the minority spin channel. The magnetic moment aligns with the Slater Pauling (SP) rule, indicating robust ferromagnetism. Mechanical analysis shows that the material is ductile in nature. Thermodynamic analysis highlights the alloy's resilience, supported by consistent trends in entropy, heat capacity, and Debye temperature.Its optical response demonstrates strong absorption in the visible and ultraviolet (UV) regions, along with pronounced dielectric and plasmonic features, suggesting potential applications in optoelectronics and refracective coatings. Furthermore, evaluations of the transport properties reveal high Seebeck coefficients and a significantly tunable figure of merit (ZT), with values approaching 0.9 across the temperature range of 300 - 1500 K, indicating excellent thermoelectric characteristics. Overall, these findings position PdCrAs as a promising multifunctional material suitable sustainable energy solutions.

cond-mat.mtrl-sci↗

arXiv2025

Thermoelectric Potential of NaVAs Half-Heusler Alloy: Insights from Ab-initio Calculations

This work presents a comprehensive investigation of the HH alloy NaVAs using first - principles methods, emphasizing its potential applications in various fields, including spintronics, thermoelectrics, and optoelectronics. We utilized density functional theory (DFT) within the full-potential linearized augmented plane wave (FLAPW) framework. Structural optimizations indicate that the alloy stabilizes in the ferromagnetic phase. Both mechanical and dynamical stability have been confirmed through analysis of elastic constants and phonon dispersion. Our calculations of the electronic band structure and density of states (DOS) reveal that NaVAs exhibits half-metallic behavior, with a spin-polarized band gap of 2.77 eV in the minority spin channel. The magnetic moment aligns with the Slater Pauling (SP) rule, demonstrating robust ferromagnetism. Mechanical analysis shows that the material is brittle in nature. The thermodynamic analysis highlights the alloy's resilience, supported by consistent trends in entropy, heat capacity, and Debye temperature. Its optical response indicates strong absorption in the visible and ultraviolet (UV) regions, along with pronounced dielectric and plasmonic features, suggesting potential for applications in optoelectronics and refective coatings. Furthermore, evaluations of the transport properties show high Seebeck coefficients and a significantly tunable figure of merit (ZT). ZT values approach 1.0 across the temperature range of 600 - 1500 K, demonstrating excellent thermoelectric characteristics. Overall, these findings position NaVAs as a promising multifunctional material suitable for advanced technological applications in green energy area.

cond-mat.mtrl-sci↗

arXiv2025

Simulating LLM training workloads for heterogeneous compute and network infrastructure

The growing demand for large-scale GPU clusters in distributed model training presents a significant barrier to innovation, particularly in model optimization, performance tuning, and system-level enhancements. To address this challenge, LLM training simulators are employed to estimate training time and guide design decisions. However, the state-of-the-art LLM training simulators assume homogeneous compute and network infrastructure. In practice, device heterogeneity is inevitable due to resource sharing in cloud environments, frequent shifts in device generations, and inherent intra-chip interconnect heterogeneity. To address the gap between state-of-the-art and practical requirements, we propose the design of a heterogeneity-aware distributed LLM simulator capable of predicting training time while enabling abstractions to specify custom configurations for device groups and device-to-parallelism mapping. We present the design requirements and challenges in building a heterogeneity-aware distributed ML training simulator, and design components such as non-uniform workload partitioning. Our initial simulation results demonstrate the impact of heterogeneity on the model computation and communication time.

cs.DC↗

arXiv2025

Beyond DNS: Unlocking the Internet of AI Agents via the NANDA Index and Verified AgentFacts

The Internet is poised to host billions to trillions of autonomous AI agents that negotiate, delegate, and migrate in milliseconds and workloads that will strain DNS-centred identity and discovery. In this paper, we describe the NANDA index architecture, which we envision as a means for discoverability, identifiability and authentication in the internet of AI agents. We present an architecture where a minimal lean index resolves to dynamic, cryptographically verifiable AgentFacts that supports multi-endpoint routing, load balancing, privacy-preserving access, and credentialed capability assertions. Our architecture design delivers five concrete guarantees: (1) A quilt-like index proposal that supports both NANDA-native agents as well as third party agents being discoverable via the index, (2) rapid global resolution for newly spawned AI agents, (3) sub-second revocation and key rotation, (4) schema-validated capability assertions, and (5) privacy-preserving discovery across organisational boundaries via verifiable, least-disclosure queries. We formalize the AgentFacts schema, specify a CRDT-based update protocol, and prototype adaptive resolvers. The result is a lightweight, horizontally scalable foundation that unlocks secure, trust-aware collaboration for the next generation of the Internet of AI agents, without abandoning existing web infrastructure.

cs.NI↗

arXiv2025

Balancing Semantic Relevance and Engagement in Related Video Recommendations

Related video recommendations commonly use collaborative filtering (CF) driven by co-engagement signals, often resulting in recommendations lacking semantic coherence and exhibiting strong popularity bias. This paper introduces a novel multi-objective retrieval framework, enhancing standard two-tower models to explicitly balance semantic relevance and user engagement. Our approach uniquely combines: (a) multi-task learning (MTL) to jointly optimize co-engagement and semantic relevance, explicitly prioritizing topical coherence; (b) fusion of multimodal content features (textual and visual embeddings) for richer semantic understanding; and (c) off-policy correction (OPC) via inverse propensity weighting to effectively mitigate popularity bias. Evaluation on industrial-scale data and a two-week live A/B test reveals our framework's efficacy. We observed significant improvements in semantic relevance (from 51% to 63% topic match rate), a reduction in popular item distribution (-13.8% popular video recommendations), and a +0.04% improvement in our topline user engagement metric. Our method successfully achieves better semantic coherence, balanced engagement, and practical scalability for real-world deployment.

cs.IR↗

arXiv2025

SHAPE -- A Spectro-Polarimeter Onboard Propulsion Module of Chandrayaan-3 Mission

SHAPE (Spectro-polarimetry of HAbitable Planet Earth) is an experiment onboard the Chandrayaan-3 Mission, designed to study the spectro-polarimetric signatures of the habitable planet Earth in the near-infrared (NIR) wavelength range (1.0 - 1.7 $μ$m). The spectro-polarimeter is the only scientific payload (experimental in nature) on the Propulsion Module (PM) of the Chandrayaan-3 mission. The instrument is a compact and lightweight spectro-polarimeter with an Acousto-Optic Tunable Filter (AOTF) at its core. The AOTF operates in the frequency range of 80 MHz to 135 MHz with a power of 0.5 - 2.0 Watts. The two output beams (e-beam and o-beam) from the AOTF are focused onto two InGaAs detectors (pixelated, 1D linear array) with the help of focusing optics. The primary (aperture) optics, with a diameter of $\sim$2 mm, collects the NIR light for input to the AOTF, defining the field of view (FOV) of 2.6$^\circ$. The payload has a mass of 4.8 kg and operates at a power of 25 Watts. This manuscript highlights some of the ground-based results, including the post-launch initial performance of the payload while orbiting around the Moon to observe Earth.

astro-ph.IM↗

arXiv2025

Generative AI for CAD Automation: Leveraging Large Language Models for 3D Modelling

Large Language Models (LLMs) are revolutionizing industries by enhancing efficiency, scalability, and innovation. This paper investigates the potential of LLMs in automating Computer-Aided Design (CAD) workflows, by integrating FreeCAD with LLM as CAD design tool. Traditional CAD processes are often complex and require specialized sketching skills, posing challenges for rapid prototyping and generative design. We propose a framework where LLMs generate initial CAD scripts from natural language descriptions, which are then executed and refined iteratively based on error feedback. Through a series of experiments with increasing complexity, we assess the effectiveness of this approach. Our findings reveal that LLMs perform well for simple to moderately complex designs but struggle with highly constrained models, necessitating multiple refinements. The study highlights the need for improved memory retrieval, adaptive prompt engineering, and hybrid AI techniques to enhance script robustness. Future directions include integrating cloud-based execution and exploring advanced LLM capabilities to further streamline CAD automation. This work underscores the transformative potential of LLMs in design workflows while identifying critical areas for future development.

cs.HC↗

arXiv2025

A Cognac Shot To Forget Bad Memories: Corrective Unlearning for Graph Neural Networks

Graph Neural Networks (GNNs) are increasingly being used for a variety of ML applications on graph data. Because graph data does not follow the independently and identically distributed (i.i.d.) assumption, adversarial manipulations or incorrect data can propagate to other data points through message passing, which deteriorates the model's performance. To allow model developers to remove the adverse effects of manipulated entities from a trained GNN, we study the recently formulated problem of Corrective Unlearning. We find that current graph unlearning methods fail to unlearn the effect of manipulations even when the whole manipulated set is known. We introduce a new graph unlearning method, Cognac, which can unlearn the effect of the manipulation set even when only 5% of it is identified. It recovers most of the performance of a strong oracle with fully corrected training data, even beating retraining from scratch without the deletion set while being 8x more efficient. We hope our work assists GNN developers in mitigating harmful effects caused by issues in real-world data, post-training. Our code is publicly available at https://github.com/cognac-gnn-unlearning/corrective-unlearning-for-gnns

cs.LG↗