arXiv subjects

Kunal Singh

Publications and source records attributed to Kunal Singh.

12 recordsLinked to original sources

arXiv2026

ProductConsistency: Improving Product Identity Preservation in Instruction-Based Image Editing via SFT and RL

Recent advances in instruction-based image editing have enabled models to perform complex visual edits from natural language instructions. However, in product-centric scenarios where preserving product features, branding, and textual elements are critical, current open and closed source models often struggle to maintain this fine-grained object identity. This issue is further compounded by the lack of datasets for instruction-based product image editing with text fidelity constraints, leaving it largely treated as an implicit capability of instruction-based image editing models. In this work, we introduce the ProductConsistency dataset which is designed to improve product-centric image editing. Our approach includes a supervised fine-tuning (SFT) dataset of 87k samples for product editing, a reinforcement learning (RL) dataset with 869 unique product images, and a new benchmark dataset, the ProductConsistency Benchmark, to allow rigorous and standardized evaluation of editing models. To guide RL training, we propose a Cyclic Consistency reward that enforces semantic preservation of product identity by using caption similarity between the original product description and captions generated from the edited image. We fine-tune both Qwen-Image-Edit-2511 and Flux.1-Kontext-dev using our dataset and demonstrate consistent improvements over baseline models in OCR and Perceptual metrics, and MLLM-based evaluations as well, indicating stronger product consistency, text rendering, and overall visual quality; with the Qwen-Image-Edit-2511 model achieving a 5x reduction in the character error rate. The code and pipeline is available at https://anonymous.4open.science/r/ProductConsistency-6FCC/README.md

cs.CV↗

arXiv2026

Time-Domain Voice Identity Morphing (TD-VIM): A Signal-Level Approach to Morphing Attacks on Speaker Verification Systems

In biometric systems, it is a common practice to associate each sample or template with a specific individual. Nevertheless, recent studies have demonstrated the feasibility of generating "morphed" biometric samples capable of matching multiple identities. These morph attacks have been recognized as potential security risks for biometric systems. However, most research on morph attacks has focused on biometric modalities that operate within the image domain, such as the face, fingerprints, and iris. In this work, we introduce Time-domain Voice Identity Morphing (TD-VIM), a novel approach for voice-based biometric morphing. This method enables the blending of voice characteristics from two distinct identities at the signal level, creating morphed samples that present a high vulnerability for speaker verification systems. Leveraging the Multilingual Audio-Visual Smartphone database, our study created four distinct morphed signals based on morphing factors and evaluated their effectiveness using a comprehensive vulnerability analysis. To assess the security impact of TD-VIM, we benchmarked our approach using the Generalized Morphing Attack Potential (G-MAP) metric, measuring attack success across two deep-learning-based Speaker Verification Systems (SVS) and one commercial system, Verispeak. Our findings indicate that the morphed voice samples achieved a high attack success rate, with G-MAP values reaching 99.40% on iPhone-11 and 99.74% on Samsung S8 in text-dependent scenarios, at a false match rate of 0.1%.

cs.SD↗

arXiv2026

An Experimental Study of Machine Learning-Based Intrusion Detection for OPC UA over Industrial Private 5G Networks

Industrial deployments increasingly rely on Open Platform Communications Unified Architecture (OPC UA) as a secure and platform-independent communication protocol, while private Fifth Generation (5G) networks provide low-latency and high-reliability connectivity for modern automation systems. However, their combination introduces new attack surfaces and traffic characteristics that remain insufficiently understood, particularly with respect to machine learning-based intrusion detection systems (ML-based IDS). This paper presents an experimental study on detecting cyberattacks against OPC UA applications operating over an operational private 5G network. Multiple attack scenarios are executed, and OPC UA traffic is captured and enriched with statistical flow-, packet-, and protocol-aware features. Several supervised ML models are trained and evaluated to distinguish benign and malicious traffic. The results demonstrate that the proposed ML-based IDS achieves high detection performance for a representative set of OPC UA-specific attack scenarios over an operational private 5G network.

cs.CR↗

arXiv2026

Mixture of Chapters: Scaling Learnt Memory in Transformers

Transformers lack an explicit architectural mechanism for storing and organizing knowledge acquired during training. We introduce learnable sparse memory banks: a set of latent tokens, randomly initialized and trained end-to-end, that transformer layers query via cross-attention to retrieve stored knowledge. To scale memory capacity without prohibitive attention costs, we propose chapter-based routing inspired by Mixture-of-Experts architectures, partitioning the memory bank into chapters and training a router to select relevant subsets per input. This enables scaling to 262K memory tokens while maintaining tractable computation. We evaluate our approach against standard transformers (in iso-FLOP settings) on pre-training and instruction fine-tuning across relevant benchmarks. Our models surpass iso-FLOP baselines suggesting scope for a new axis of scaling, demonstrating that explicit associative memory provides complementary capacity to what is captured implicitly in model parameters. Additionally, we observe improved knowledge retention under continued training, with robustness to forgetting when transitioning between training phases (e.g., pretraining to instruction fine-tuning).

cs.LG↗

arXiv2026

Non-equilibrium dynamics of the disordered Power of Two model

Motivated by recent experimental realizations of programmable spin models with long-range interactions, we investigate the non-equilibrium dynamics of the Power-of-Two (PWR2) model. This model consists of sparse long-range couplings between spin-$1/2$ objects separated by $d = 2^n$. In the absence of disorder, the system exhibits rapid scrambling and fast thermalization. We explore the impact of disorder in this system by analyzing the time evolution of the survival probability, half-chain entanglement entropy, and out-of-time-ordered correlators (OTOCs). We find that sufficiently strong disorder suppresses information spreading and induces localization. Remarkably, in the strong-disorder regime, the OTOCs display a non-monotonic spatial profile arising from the intrinsic nonlocality of the interactions, signaling qualitatively distinct scrambling dynamics compared to conventional long-range interacting systems. To characterize the localization transition, we extract the critical disorder strength $h_c$ from the spectral statistics and the eigenstate entanglement. We observe that $h_c$ increases with system size. Furthermore, at a fixed disorder strength, the eigenstate-averaged entanglement entropy increases with system size, while the inverse participation ratio decreases, indicating enhanced delocalization at larger sizes. These results collectively suggest that the PWR2 model remains ergodic in the thermodynamic limit for any finite disorder strength.

cond-mat.dis-nn↗

arXiv2026

Making Databases Searchable with Deep Context

Databases are the most critical assets for enterprises, and yet they remain largely inaccessible to people who make the most important decisions. In this paper, we describe the Tursio search platform that builds an abstraction layer, aka semantic knowledge graph, over the underlying databases to make them searchable in natural language. Tursio infuses large language models (LLMs) into every part of the query processing stack, including data modeling, query compilation, query planning, and result reasoning. This allows Tursio to process natural language queries systematically using techniques from traditional query planning and rewriting, rather than black-box memorization. We describe the architecture of Tursio in detail and present a comprehensive evaluation on production workloads, and synthetic and realistic benchmarks. Our results show that Tursio achieves high accuracy while being efficient and scalable, making databases truly searchable for non-expert users.

cs.DB↗

arXiv2025

Fathom-DeepResearch: Unlocking Long Horizon Information Retrieval and Synthesis for SLMs

Tool-integrated reasoning has emerged as a key focus for enabling agentic applications. Among these, DeepResearch Agents have gained significant attention for their strong performance on complex, open-ended information-seeking tasks. We introduce Fathom-DeepResearch, an agentic system composed of two specialized models. The first is Fathom-Search-4B, a DeepSearch model trained from Qwen3-4B and optimized for evidence-based investigation through live web search and targeted webpage querying. Its training combines three advances: (i) DUETQA, a 5K-sample dataset generated via multi-agent self-play that enforces strict web-search dependence and heterogeneous source grounding; (ii) RAPO, a zero-overhead extension of GRPO that stabilizes multi-turn Reinforcement Learning with Verifiable Rewards through curriculum pruning, reward-aware advantage scaling, and per-prompt replay buffers; and (iii) a steerable step-level reward that classifies each tool call by cognitive behavior and marginal utility, enabling explicit control over search trajectory breadth, depth, and horizon. These improvements enable reliable extension of tool-calling beyond 20 calls when warranted. The second is Fathom-Synthesizer-4B, trained from Qwen3-4B, which converts multi-turn DeepSearch traces into structured, citation-dense DeepResearch Reports for comprehensive synthesis. Evaluated on DeepSearch benchmarks (SimpleQA, FRAMES, WebWalker, Seal0, MuSiQue) and DeepResearch-Bench, the system achieves state-of-the-art performance in the open-weights category while demonstrating strong generalization to diverse reasoning tasks including HLE, AIME-25, GPQA-Diamond, and MedQA.

cs.AI↗

arXiv2025

SBSC: Step-By-Step Coding for Improving Mathematical Olympiad Performance

We propose Step-by-Step Coding (SBSC): a multi-turn math reasoning framework that enables Large Language Models (LLMs) to generate sequence of programs for solving Olympiad level math problems. At each step/turn, by leveraging the code execution outputs and programs of previous steps, the model generates the next sub-task and the corresponding program to solve it. This way, SBSC, sequentially navigates to reach the final answer. SBSC allows more granular, flexible and precise approach to problem-solving compared to existing methods. Extensive experiments highlight the effectiveness of SBSC in tackling competition and Olympiad-level math problems. For Claude-3.5-Sonnet, we observe SBSC (greedy decoding) surpasses existing state-of-the-art (SOTA) program generation based reasoning strategies by absolute 10.7% on AMC12, 8% on AIME and 12.6% on MathOdyssey. Given SBSC is multi-turn in nature, we also benchmark SBSC's greedy decoding against self-consistency decoding results of existing SOTA math reasoning strategies and observe performance gain by absolute 6.2% on AMC, 6.7% on AIME and 7.4% on MathOdyssey.

cs.AI↗

arXiv2025

TRISHUL: Towards Region Identification and Screen Hierarchy Understanding for Large VLM based GUI Agents

Recent advancements in Large Vision Language Models (LVLMs) have enabled the development of LVLM-based Graphical User Interface (GUI) agents under various paradigms. Training-based approaches, such as CogAgent and SeeClick, struggle with cross-dataset and cross-platform generalization due to their reliance on dataset-specific training. Generalist LVLMs, such as GPT-4V, employ Set-of-Marks (SoM) for action grounding, but obtaining SoM labels requires metadata like HTML source, which is not consistently available across platforms. Moreover, existing methods often specialize in singular GUI tasks rather than achieving comprehensive GUI understanding. To address these limitations, we introduce TRISHUL, a novel, training-free agentic framework that enhances generalist LVLMs for holistic GUI comprehension. Unlike prior works that focus on either action grounding (mapping instructions to GUI elements) or GUI referring (describing GUI elements given a location), TRISHUL seamlessly integrates both. At its core, TRISHUL employs Hierarchical Screen Parsing (HSP) and the Spatially Enhanced Element Description (SEED) module, which work synergistically to provide multi-granular, spatially, and semantically enriched representations of GUI elements. Our results demonstrate TRISHUL's superior performance in action grounding across the ScreenSpot, VisualWebBench, AITW, and Mind2Web datasets. Additionally, for GUI referring, TRISHUL surpasses the ToL agent on the ScreenPR benchmark, setting a new standard for robust and adaptable GUI comprehension.

cs.CV↗

arXiv2023

Gradual Verification for Smart Contracts

Blockchains facilitate secure resource transactions through smart contracts, yet these digital agreements are prone to vulnerabilities, particularly when interacting with external contracts, leading to substantial monetary losses. Traditional verification techniques fall short in providing comprehensive security assurances, especially against re-entrancy attacks, due to the unavailable implementations of external contracts. This paper introduces an incremental approach: gradual verification. We combine static and dynamic verification techniques to enhance security, guarantee soundness and flexibility, and optimize resource usage in smart contract interactions. By implementing a prototype for gradually verifying Algorand smart contracts via the pyTEAL language, we demonstrate the effectiveness of our approach, contributing to the safe and efficient execution of smart contracts.

cs.CR↗

arXiv2019

Aspect-Sentiment Embeddings for Company Profiling and Employee Opinion Mining

With the multitude of companies and organizations abound today, ranking them and choosing one out of the many is a difficult and cumbersome task. Although there are many available metrics that rank companies, there is an inherent need for a generalized metric that takes into account the different aspects that constitute employee opinions of the companies. In this work, we aim to overcome the aforementioned problem by generating aspect-sentiment based embedding for the companies by looking into reliable employee reviews of them. We created a comprehensive dataset of company reviews from the famous website Glassdoor.com and employed a novel ensemble approach to perform aspect-level sentiment analysis. Although a relevant amount of work has been done on reviews centered on subjects like movies, music, etc., this work is the first of its kind. We also provide several insights from the collated embeddings, thus helping users gain a better understanding of their options as well as select companies using customized preferences.

cs.CL↗

arXiv2019

TEduChain: A Platform for Crowdsourcing Tertiary Education Fund using Blockchain Technology

Blockchain is an emerging technology framework for creating and storing transaction in distributed ledgers with a high degree of security and reliability. In this paper we present a blockchain-based platform to create and store contracts in between students and their higher education sponsors. The sponsorship might be in any form, such as scholarship, donation or loan. The fund will be arranged and managed by a group of competitive agents (Fundraisers) who will hold the distributed ledgers and act as miners in the blockchain network.

cs.CY↗