arXiv subjects

Avinash Kumar

Publications and source records attributed to Avinash Kumar.

At least 19 recordsLinked to original sources

arXiv2026

Spectroscopic redshifts of selected flat-spectrum radio sources I

We present the first results of a spectroscopic campaign carried out as part of the Search for Milli-Lenses (SMILE) program, which aims to constrain the prevalence of gravitational lens systems on milli-arcsecond angular scales (milli-lenses) using high-resolution Very Long Baseline Interferometry (VLBI) imaging. The SMILE parent sample contains ~ 5000 radio-loud active galaxies, selected as a flux-limited, complete subsample of CLASS (The Cosmic Lens All-Sky Survey) sources. We compiled redshift information for the full sample from multiple literature and catalog sources and found that 491 sources have no available redshift estimate, either spectroscopic or photometric. A further 948 sources have only photometric redshifts, many of which show substantial discrepancies between catalogs. Reliable redshifts are essential for VLBI radio-source studies because they convert angular measurements into physical linear scales, enable estimates of intrinsic luminosities and jet kinematics, and allow robust cosmological and population studies. To address this key limitation for lensing and population studies, we initiated a dedicated spectroscopic campaign to secure reliable redshifts for as many targets as possible. This paper focuses on the brightest sources in the SMILE sample. We report newly determined spectroscopic redshifts for 6 targets out of 15 observed with the Skinakas 1.3 m telescope.

astro-ph.CO↗

arXiv2026

Context Graphs for Proactive Enterprise Agents

Retrieval-Augmented Generation (RAG) and agentic frameworks have advanced enterprise AI considerably, yet agents remain fundamentally reactive: they wait for a human query before acting. This paper argues that genuine enterprise productivity gains require proactive agents: systems that surface relevant, actionable information to workers before they ask. We propose the Context Graph, a live relational data structure that models enterprise entities, their relationships, and state transitions over time. Built on this graph, we define a Delta Detection Engine that continuously monitors state changes, a Proactivity Scorer that ranks candidate insights by urgency, relevance, and persona-fit, and a Surfacing Layer powered by an LLM that delivers ranked notifications with grounded explanations. We formalize each component, derive a unified Proactivity Score function, and provide a complete end-to-end Python implementation using NetworkX and the Anthropic Claude API. Evaluation across three generic enterprise case studies (contract lifecycle management, engineering incident response, and sales pipeline hygiene) demonstrates that context-graph-driven proactivity achieves Precision@5 of 0.83, a false positive rate of 0.11, and reduces mean time to surface from 47 minutes (reactive baseline) to under 30 second.

cs.AI↗

arXiv2026

GEM: GPU-Variability-Aware Expert to GPU Mapping for MoE Systems

Mixture-of-Expert (MoE) models enable efficient inference by employing smaller experts and activating only a subset of them per token. MoE serving engines distribute experts across multiple GPUs and route tokens to appropriate GPUs at inference time based on experts activated. They process tokens in lock-step fashion, where tokens within a batch must finish processing before proceeding to the next layer. This synchronization barrier acts as a critical bottleneck because the performance of MoE models is limited by the straggler GPU that finishes last. Stragglers emerge when too many heavily used experts are placed on the same GPU or the slowest GPU. While prior works place experts that balance token loads across GPUs, they all overlook GPU variability and often place highly used experts on the slowest GPUs. We propose GEM, GPU-variability-aware Expert Mapping, a framework for GPU variability-aware expert to GPU mapping for MoE models. GEM exploits two insights. First, we must place experts such that each GPU receives non-uniform token loads based on their variability and they all finish processing a layer at about the same time. Our studies show that there are two types of experts: consistent that are used most of the time and temporal that are often used together for the remaining time. Our second insight is that we must place simultaneously used consistent and temporal experts on different GPUs and avoid placing them on slower GPUs to reduce slowdown. GEM gathers the variability profile of GPUs for each model and task and uses the token load distributions per task to map experts to GPUs. Our experiments show that GEM improves end-to-end latency by 7.9% on average and by up to 16.5% compared to the baseline.

cs.DC↗

arXiv2026

Test-Time Speculation

Speculative decoding accelerates LLM inference by using a fast draft model to generate tokens and a more accurate target model to verify them. Its performance depends on the $\textit{acceptance length}$, or number of draft tokens accepted by the target. Our studies show that the acceptance length of even state-of-the-art speculators, like DFlash, EAGLE-3 and PARD degrade with generation length, reaching values close to 1 (i.e. no speedup) within just a few thousand output tokens, making speculators ineffective for long-response tasks. Acceptance lengths decline because most speculators are trained offline on short sequences, but are forced to match the target model on much longer outputs at inference, well beyond their training distribution. To address this issue, we propose $\textit{Test-Time Speculation (TTS)}$, an online distillation approach that continuously adapts the speculator at test-time. TTS leverages the key insight that the token verification step already invokes the target model for each draft token, providing the training signal needed to adapt the draft at no additional cost. Treating the draft as the student and the target as a teacher, TTS adjusts the draft over several speculation rounds, with each update improving the draft's accuracy as generation proceeds. Our results across multiple models from the Qwen-3, Qwen-3.5, and Llama3.1 families show that TTS improves acceptance lengths over state-of-the-art speculators by up to $72\%$ and $41\%$ on average, with the benefits scaling with increased generation lengths.

cs.CL↗

arXiv2026

Health+: Empowering Individuals via Unifying Health Data

Managing personal health data is a challenge in today's fragmented and institution-centric healthcare ecosystem. Individuals often lack meaningful control over their medical records, which are scattered across incompatible systems and formats. This vision paper presents Health+, a user-centric, multimodal health data management system that empowers individuals (including those with limited technical expertise) to upload, query, and share their data across modalities (e.g., text, images, reports). Rather than aiming for institutional overhaul, Health+ emphasizes individual agency by providing intuitive interfaces and intelligent recommendations for data access and sharing. At the system level, it tackles the complexity of storing, integrating, and securing heterogeneous health records, ensuring both efficiency and privacy. By unifying multimodal data and prioritizing patients, Health+ lays the foundation for a more connected, interpretable, and user-controlled health information ecosystem.

cs.MM↗

arXiv2026

Recall, Risk, and Governance in Automated Proposal Screening for Research Funding: Evidence from a National Funding Programme

Research funding agencies are increasingly exploring automated tools to support early-stage proposal screening. Recent advances in large language models (LLMs) have generated optimism regarding their use for text-based evaluation, yet their institutional suitability for high-stakes screening decisions remains underexplored. In particular, there is limited empirical evidence on how automated screening systems perform when evaluated against institutional error costs. This study compares two automated approaches for proposal screening against the priorities of a national funding call: A transparent, rule-based method using term frequency-inverse document frequency (TF-IDF) with domain-specific keyword engineering, and a semantic classification approach based on a large language model. Using selection committee decisions as ground truth for 959 proposals, we evaluate performance with particular attention to error structure. The results show that the TF-IDF-based approach outperforms the LLM-based system across standard metrics, achieving substantially higher recall (78.95\% vs 45.82\%) and producing far fewer false negatives (68 vs 175). The LLM-based system excludes more than half of the proposals ultimately selected by the committee. While false positives can be corrected through subsequent peer review, false negatives represent an irrecoverable exclusion from expert evaluation. By foregrounding error asymmetry and institutional context, this study demonstrates that the suitability of automated screening systems depends not on model sophistication alone, but on how their error profiles, transparency, and auditability align with research evaluation practice. These findings suggest that evaluation design and error tolerance should guide the use of AI-assisted screening tools in research funding more broadly.

cs.DL↗

arXiv2025

Latent Sculpting for Zero-Shot Generalization: A Manifold Learning Approach to Out-of-Distribution Anomaly Detection

Detecting previously unseen attacks remains a major challenge for machine learning-based intrusion detection systems. Deep models trained on network traffic often achieve high accuracy on known attacks but fail under distributional shift because their decision boundaries are tightly coupled to the training data distribution. We introduce Latent Sculpting, a two-stage anomaly detection framework that improves robustness by explicitly structuring the latent representation before density estimation. The first stage trains a Transformer-based tabular encoder using a novel Binary Latent Sculpting loss, which encourages benign traffic to form a compact latent cluster while enforcing separation from anomalous patterns. The second stage fits a Masked Autoregressive Flow to the resulting latent space to produce calibrated probabilistic anomaly scores. Under a strict zero-shot evaluation protocol on the CIC-IDS-2017 benchmark, Stage 1 attains an F1-score of 0.98 on known attacks, while Stage 2 -- evaluated at the balanced threshold (85th-percentile) -- achieves a zero-shot OOD F1-score of 0.867 and AUROC of 0.913. The model successfully detects difficult distribution shifts including stealthy infiltration attacks (78.7% recall, peaking at 97.2%) and low-volume DoS variants (>94% recall), scenarios where conventional approaches often fail. Our results suggest that explicitly separating latent geometry learning from density modeling provides a stable approach for detecting zero-day cyber threats.

cs.LG↗

arXiv2025

QuArch: A Benchmark for Evaluating LLM Reasoning in Computer Architecture

The field of computer architecture, which bridges high-level software abstractions and low-level hardware implementations, remains absent from current large language model (LLM) evaluations. To this end, we present QuArch (pronounced 'quark'), the first benchmark designed to facilitate the development and evaluation of LLM knowledge and reasoning capabilities specifically in computer architecture. QuArch v1.0 provides a comprehensive collection of 2,671 expert-validated question-answer (QA) pairs covering various aspects of computer architecture, including processor design, memory systems, and interconnection networks. Our evaluation reveals that while frontier models possess domain-specific knowledge, they struggle with skills that require higher-order thinking in computer architecture. Frontier model accuracies vary widely (from 34% to 73%) on these advanced questions, highlighting persistent gaps in architectural reasoning across analysis, design, and implementation QAs. Furthermore, via fine-tuning we find that QuArch can translate to improved performance on a realistic memory hierarchy design task, resulting in up to 1.99x more area-efficient solutions and up to 40% more viable solutions overall. By holistically assessing fundamental skills, QuArch provides a foundation for building and measuring LLM capabilities that can accelerate innovation in computing systems. The QuArch benchmark and leaderboard are publicly available at: https://quarch.ai/.

cs.AR↗

arXiv2025

HiSpec: Hierarchical Speculative Decoding for LLMs

Speculative decoding accelerates LLM inference by using a smaller draft model to speculate tokens that a larger target model verifies. Verification is often the bottleneck (e.g. verification is $4\times$ slower than token generation when a 3B model speculates for a 70B target model), but most prior works focus only on accelerating drafting. $\textit{``Intermediate"}$ verification reduces verification time by discarding inaccurate draft tokens early, but existing methods incur substantial training overheads in incorporating the intermediate verifier, increase the memory footprint to orchestrate the intermediate verification step, and compromise accuracy by relying on approximate heuristics. We propose $\underline{\textit{Hi}}\textit{erarchical }\underline{\textit{Spec}}\textit{ulative Decoding (HiSpec)}$, a framework for high-throughput speculative decoding that exploits $\textit{early-exit (EE) models}$ for low-overhead intermediate verification. EE models allow tokens to exit early by skipping layer traversal and are explicitly trained so that hidden states at selected layers can be interpreted, making them uniquely suited for intermediate verification without drastically increasing compute and memory overheads. To improve resource-efficiency even further, we design a methodology that enables HiSpec to re-use key-value caches and hidden states between the draft, intermediate verifier, and target models. To maintain accuracy, HiSpec periodically validates the draft tokens accepted by the intermediate verifier against the target model. Our evaluations using various representative benchmarks and models show that HiSpec improves throughput by 1.28$\times$ on average and by up to 2.01$\times$ compared to the baseline single-layer speculation without compromising accuracy.

cs.CL↗

arXiv2025

VIPCALs: A fully automated calibration pipeline for very long baseline interferometry data

Very long baseline interferometry (VLBI) is a powerful technique that can achieve sub-milliarcsecond resolution. However, it requires complex and often manual post-correlation calibration to correct for instrumental, geometric, and propagation-related errors. Unlike connected-element interferometers, VLBI arrays typically provide raw visibilities rather than science-ready data, and existing pipelines are largely semi-automated and reliant on user supervision. We present VIPCALs, a fully automated, end-to-end calibration pipeline for continuum VLBI data that operates without human intervention or prior knowledge of the dataset. Designed for scalability to thousands of sources and heterogeneous archival observations, VIPCALs addresses the needs of initiatives such as the Search for Milli-Lenses (SMILE) project. Implemented in Python using ParselTongue, VIPCALs reproduces the standard AIPS calibration workflow in a fully unsupervised mode. Besides the usual calibration tasks, the pipeline also performs automatic reference antenna selection, calibrator identification, and generates diagnostic outputs for inspection. We validated it on a representative sample of Very Long Baseline Array (VLBA) data corresponding to 1000 sources from the SMILE project. VIPCALs successfully calibrated observations of 955 of the test sources across multiple frequency bands. Over 91% of the calibrated datasets achieved successful fringe fitting on target in at least half of the solutions attempted. The median ratio of calibrated visibilities to initial total visibilities was 0.87. The average processing time was below 10 minutes per dataset, demonstrating both efficiency and scalability. VIPCALs enables robust, reproducible, and fully automated calibration of VLBI continuum data, significantly lowering the entry barrier for VLBI science and making large-scale projects like SMILE feasible.

astro-ph.IM↗

arXiv2025

A Study on the Refining Handwritten Font by Mixing Font Styles

Handwritten fonts have a distinct expressive character, but they are often difficult to read due to unclear or inconsistent handwriting. FontFusionGAN (FFGAN) is a novel method for improving handwritten fonts by combining them with printed fonts. Our method implements generative adversarial network (GAN) to generate font that mix the desirable features of handwritten and printed fonts. By training the GAN on a dataset of handwritten and printed fonts, it can generate legible and visually appealing font images. We apply our method to a dataset of handwritten fonts and demonstrate that it significantly enhances the readability of the original fonts while preserving their unique aesthetic. Our method has the potential to improve the readability of handwritten fonts, which would be helpful for a variety of applications including document creation, letter writing, and assisting individuals with reading and writing difficulties. In addition to addressing the difficulties of font creation for languages with complex character sets, our method is applicable to other text-image-related tasks, such as font attribute control and multilingual font style transfer.

cs.CV↗

arXiv2025

Text-Conditioned Diffusion Model for High-Fidelity Korean Font Generation

Automatic font generation (AFG) is the process of creating a new font using only a few examples of the style images. Generating fonts for complex languages like Korean and Chinese, particularly in handwritten styles, presents significant challenges. Traditional AFGs, like Generative adversarial networks (GANs) and Variational Auto-Encoders (VAEs), are usually unstable during training and often face mode collapse problems. They also struggle to capture fine details within font images. To address these problems, we present a diffusion-based AFG method which generates high-quality, diverse Korean font images using only a single reference image, focusing on handwritten and printed styles. Our approach refines noisy images incrementally, ensuring stable training and visually appealing results. A key innovation is our text encoder, which processes phonetic representations to generate accurate and contextually correct characters, even for unseen characters. We used a pre-trained style encoder from DG FONT to effectively and accurately encode the style images. To further enhance the generation quality, we used perceptual loss that guides the model to focus on the global style of generated images. Experimental results on over 2000 Korean characters demonstrate that our model consistently generates accurate and detailed font images and outperforms benchmark methods, making it a reliable tool for generating authentic Korean fonts across different styles.

cs.CV↗

arXiv2025

HELIOS: Adaptive Model And Early-Exit Selection for Efficient LLM Inference Serving

Early-Exit Large Language Models (EE-LLMs) enable high throughput inference by allowing tokens to exit early at intermediate layers. However, their throughput is limited by the computational and memory savings. Existing EE-LLM frameworks rely on a single model and therefore, their token generation latencies are bottlenecked by tokens that do not exit early and traverse additional layers. Moreover, early exits are only known at runtime and depend on the request. Therefore, these frameworks load the weights of all model layers even though large portions remain unused when tokens exit early. The lack of memory savings limit us from scaling the batch sizes. We propose $\textit{HELIOS}$, a framework that improves both token generation latency and batch sizes to enable high-throughput in EE-LLMs. HELIOS exploits two insights. $\textit{First}$, early exits are often complimentary across models, tokens that do not exit early on one model often take an early-exit on another. HELIOS employs multiple models and dynamically switches between them to collectively maximize the number of tokens that exit early, and minimize token generation latencies. $\textit{Second}$, even when a predicted token does not exit early due to poor confidence, it often remains unchanged even after additional layer traversal. HELIOS greedily allows such tokens to exit early and only loads the weights of the most likely to be used layers, yielding memory savings which is then re-purposed to increase batch sizes. HELIOS employs real-time profiling to accurately identify the early-exit distributions, and adaptively switches between models by tracking tokens in real-time to minimize the performance degradation caused by greedy model loading and exiting. Our evaluations show that HELIOS achieves $1.48\times$ higher throughput and $15.14\times$ larger batch size compared to existing EE-LLM frameworks.

cs.CL↗

arXiv2025

Context Switching for Secure Multi-programming of Near-Term Quantum Computers

Multi-programming quantum computers improve device utilization and throughput. However, crosstalk from concurrent two-qubit CNOT gates poses security risks, compromising the fidelity and output of co-running victim programs. We design Zero Knowledge Tampering Attacks (ZKTAs), using which attackers can exploit crosstalk without knowledge of the hardware error profile. ZKTAs can alter victim program outputs in 40% of cases on commercial systems. We identify that ZKTAs succeed because the attacker's program consistently runs with the same victim program in a fixed context. To mitigate this, we propose QONTEXTS: a context-switching technique that defends against ZKTAs by running programs across multiple contexts, each handling only a subset of trials. QONTEXTS uses multi-programming with frequent context switching while identifying a unique set of programs for each context. This helps limit only a fraction of execution to ZKTAs. We enhance QONTEXTS with attack detection capabilities that compare the distributions from different contexts against each other to identify noisy contexts executed with ZKTAs. Our evaluations on real IBMQ systems show that QONTEXTS increases program resilience by three orders of magnitude and fidelity by 1.33$\times$ on average. Moreover, QONTEXTS improves throughput by 2$\times$, advancing security in multi-programmed environments.

cs.CR↗

arXiv2025

Choosing Augmentation Parameters in OSQP- A New Approach based on Conjugate Directions

This work proposes a new method to select the augmentation parameters in the operator splitting quadratic program (OSQP) algorithm so as to reduce the computation time of overall algorithm. The selection is based upon the information of conjugate directions of the coefficient matrix of a linear system of equations present in the algorithm. This selection makes it possible to cache these conjugate directions, instead of computing them at each iteration, resulting in faster computation of the solution of the linear system thus reducing the overall computation time. This reduction is demonstrated by a numerical example.

math.OC↗

arXiv2025

Dialogue Without Limits: Constant-Sized KV Caches for Extended Responses in LLMs

Autoregressive Transformers rely on Key-Value (KV) caching to accelerate inference. However, the linear growth of the KV cache with context length leads to excessive memory consumption and bandwidth constraints. This bottleneck is particularly problematic in real-time applications -- such as chatbots and interactive assistants -- where low latency and high memory efficiency are critical. Existing methods drop distant tokens or compress states in a lossy manner, sacrificing accuracy by discarding vital context or introducing bias. We propose MorphKV, an inference-time technique that maintains a constant-sized KV cache while preserving accuracy. MorphKV balances long-range dependencies and local coherence during text generation. It eliminates early-token bias while retaining high-fidelity context by adaptively ranking tokens through correlation-aware selection. Unlike heuristic retention or lossy compression, MorphKV iteratively refines the KV cache via lightweight updates guided by attention patterns of recent tokens. This approach captures inter-token correlation with greater accuracy, crucial for tasks like content creation and code generation. Our studies on long-response tasks show 52.9$\%$ memory savings and 18.2$\%$ higher accuracy on average compared to state-of-the-art prior works, enabling efficient real-world deployment.

cs.CL↗

arXiv2025

Lyapunov-like Stability Inequality with an Asymmetric Matrix and Application to Suboptimal LQ Control Design

The Lyapunov inequality is an indispensable tool for stability analysis in linear control theory. It provides a necessary and sufficient condition for the stability of an autonomous linear-time invariant system in terms of the existence of a symmetric positive-definite Lyapunov matrix. This work proposes a new variant of this inequality in which the constituent Lyapunov matrix is allowed to be asymmetric. After analysing the properties of the proposed inequality for a class of matrices, we derive new results for the stabilisation of linear systems. Subsequently, we utilize the developed results to obtain sufficient conditions for the suboptimal linear quadratic control design problem wherein addition to having an asymmetric Lyapunov matrix, which serves as a design matrix for this problem, we provide a characterization of the cost associated with the computed stabilizing suboptimal control laws. This characterization is done by deriving an expression for the upper bound on cost in terms of the initial conditions of the system. We demonstrate the applicability of the proposed results using two numerical examples -- one for suboptimal control design for a linear time-invariant system and another for the consensus (state-agreement) protocol design for a multi-agent system, wherein we see how the asymmetry of the Lyapunov (design) matrix emerges as an inherent requirement for the problem.

math.OC↗

arXiv2025

READ: Reinforcement-based Adversarial Learning for Text Classification with Limited Labeled Data

Pre-trained transformer models such as BERT have shown massive gains across many text classification tasks. However, these models usually need enormous labeled data to achieve impressive performances. Obtaining labeled data is often expensive and time-consuming, whereas collecting unlabeled data using some heuristics is relatively much cheaper for any task. Therefore, this paper proposes a method that encapsulates reinforcement learning-based text generation and semi-supervised adversarial learning approaches in a novel way to improve the model's performance. Our method READ, Reinforcement-based Adversarial learning, utilizes an unlabeled dataset to generate diverse synthetic text through reinforcement learning, improving the model's generalization capability using adversarial learning. Our experimental results show that READ outperforms the existing state-of-art methods on multiple datasets.

cs.CL↗