arXiv subjects

Prathosh A P

Publications and source records attributed to Prathosh A P.

15 recordsLinked to original sources

arXiv2026

Spectral Query-Key Product Weight Steering for Training-Free VLM Hallucination Mitigation

Vision-language models (VLMs) often generate fluent but visually unsupported descriptions, especially by mentioning objects absent from the image. We propose QK Product Steering, a data-free, training-free, and zero-inference-cost weight edit for reducing object hallucination. The method directly edits the per-head query-key product, the operator that produces pre-softmax attention logits, by suppressing a small number of dominant singular modes in selected middle layers. The edited product is then mapped back to the query weights through a closed-form query-only update while keeping shared key weights fixed, making the edit compatible with grouped-query attention. We further decompose the QK product into symmetric and antisymmetric components to distinguish mutual content-similarity patterns from directional attention patterns. Across three GQA-based VLMs, QK Product Steering achieves an average relative CHAIR$_s$ reduction of $4.0\%$, while matched random-mode controls show negligible change. Interpretability ablations show that the hallucination signal is specific to dominant QK modes and is primarily localized to the symmetric mutual-attention channel. Overall, QK Product Steering offers a simple alternative to decoding-time mitigation, requiring no additional data, fine-tuning, or inference-time overhead while largely preserving general multimodal capability.

cs.CV↗

arXiv2026

Vagdhenu: A Vrutta (Meter) Aware Shloka-to-Chant (TTS) System for Sanskrit

We present Vagdhenu, a vrutta (meter) aware shloka-to-chant system for Sanskrit: a text-to-speech system that maps a metrical verse to its chanted parayana recitation at high fidelity. This is an experience report, not a new architecture. We take an off-the-shelf flow-matching TTS backbone and a large-scale neural vocoder, and add the components a faithful Sanskrit chant pipeline needs: a frontend that routes Sanskrit through Kannada orthography to avoid the Hindi-style schwa deletion that Devanagari triggers in Indic models; a frontend that obeys subtle Sanskrit phonology (visarga sandhi with its jihvamuliya and upadhmaniya allophones, the aspiration contrast of alpaprana and mahaprana, and the dental, retroflex, and palatal sibilants kept distinct); and a vrutta-aware mechanism that detects the meter and picks an exactly matched reference under a half-reference rule. We report a negative result that shaped the system: in a self-infilling flow-matching backbone, a text-side prosody conditioner is architecturally inert, because the model recovers pitch from the context mel and the embedding gets no gradient; the reference clip and a voice-steering retrain are the only working prosody levers. We also report a comparative lineage across four families (StyleTTS2, VITS2, Matcha-TTS, and the flow-matching backbone), where each earlier family hit a ceiling on conjuncts or prosody that a five-hour clone cleared at an expert MOS near 4.6. The system shipped two deployments: a 32-chapter, 5183-verse video corpus (about 17.5 hours) and an audio app covering about 18000 verses across 12 books. We release the frontend, inference and training code, weights, a single-speaker chant dataset, and an interactive demo.

cs.CL↗

arXiv2026

Graph Mamba Operator: A Latent Simulator for Interacting Particle Systems

Modeling interacting dynamical systems requires capturing spatial interactions alongside long-range temporal dependencies. Graph neural networks (GNNs) provide a natural representation but typically rely on autoregressive rollouts and treat spatial and temporal dynamics separately, leading to error accumulation over long horizons. Existing approaches also focus on local interactions and short temporal contexts, limiting their ability to capture multi-hop dependencies and global structure. We introduce the Graph Mamba Operator (GraMO), a latent-space simulator that integrates state-space models with graph-based interaction learning. In contrast to prior work that sequences nodes or applies spatial and temporal updates in separate stages, GraMO couples graph-based interactions and temporal state updates within a single recurrence. The update is linear in the latent state, with input-dependent coefficients that adapt across regimes. We evaluate GraMO on N-body systems, motion capture, and robotics datasets, achieving the lowest error across benchmarks and the largest gains in long-horizon prediction.

cs.LG↗

arXiv2026

BiMol-Diff: A Unified Diffusion Framework for Molecular Generation and Captioning

Bridging molecular structures and natural language is essential for controllable design. Autoregressive models struggle with long-range dependencies, while standard diffusion processes apply uniform corruption across positions, which can distort structurally informative tokens. We present BiMol-Diff, a unified diffusion framework for the paired tasks of text-conditioned molecule generation and molecule captioning. Our key component is a token-aware noise schedule that assigns position-dependent corruption based on token recovery difficulty, preserving harder-to-recover substructures during the forward process. On ChEBI-20 and M3-20M, BiMol-Diff improves molecule reconstruction with a 15.4% relative gain in Exact Match and achieves strong captioning results, attaining best BLEU and BERTScore among compared baselines. These results indicate token-aware noising improves fidelity in molecular structure-language modelling.

cs.CL↗

arXiv2026

Factual and Edit-Sensitive Graph-to-Sequence Generation via Graph-Aware Adaptive Noising

Fine-tuned autoregressive models for graph-to-sequence generation (G2S) often struggle with factual grounding and edit sensitivity. To tackle these issues, we propose a non-autoregressive diffusion framework that generates text by iterative refinement conditioned on an input graph, named as Diffusion Language Model for Graphs (DLM4G). By aligning graph components (entities/relations) with their corresponding sequence tokens, DLM4G employs an adaptive noising strategy. The proposed strategy uses per-token denoising error as a signal to adaptively modulate noise on entity and relation tokens, improving preservation of graph structure and enabling localized updates under graph edits. Evaluated on three datasets, DLM4G consistently outperforms competitive G2S diffusion baselines trained on identical splits across both surface-form and embedding-based metrics. DLM4G further exceeds fine-tuned autoregressive baselines up to 12x larger (e.g., T5-Large) and is competitive with zero-shot LLM transfer baselines up to 127x larger. Relative to the strongest fine-tuned PLM baseline, DLM4G improves factual grounding (FGT@0.5) by +5.16% and edit sensitivity (ESR) by +7.9%; compared to the best diffusion baseline, it yields gains of +3.75% in FGT@0.5 and +23.6% in ESR. We additionally demonstrate applicability beyond textual graphs through experiments on molecule captioning, indicating the method's generality for scientific G2S generation.

cs.CL↗

arXiv2026

Interpretable Discovery of One-parameter Subgroups: A Modular Framework for Elliptical, Hyperbolic, and Parabolic Symmetries

We propose a modular, data-driven framework for jointly learning unknown functional mappings and discovering the underlying one-parameter symmetry subgroup governing the data. Unlike conventional geometric deep learning methods that assume known symmetries, our approach identifies the relevant continuous subgroup directly from data. We consider the broad class of one-parameter subgroups, which admit a canonical geometric classification into three regimes: elliptical, hyperbolic, and parabolic. Given an assumed regime, our framework instantiates a corresponding symmetry discovery architecture with invariant and equivariant representation layers structured according to the Lie algebra of the subgroup, and learns the exact generator parameters end-to-end from data. This yields models whose invariance or equivariance is guaranteed by construction and admits formal proofs, enabling symmetry to be explicitly traced to identifiable components of the architecture. The approach is applicable to one-parameter subgroups of a wide range of matrix Lie groups, including $SO(n)$, $SL(n)$, and the Lorentz group. Experiments on synthetic and real-world systems, including moment of inertia prediction, double-pendulum dynamics, and high-energy \textit{Top Quark Tagging}, demonstrate accurate subgroup recovery and strong predictive performance across both compact and non-compact regimes.

cs.LG↗

arXiv2025

Learning Equivariant Functions via Quadratic Forms

In this study, we introduce a method for learning group (known or unknown) equivariant functions by learning the associated quadratic form $x^T A x$ corresponding to the group from the data. Certain groups, known as orthogonal groups, preserve a specific quadratic form, and we leverage this property to uncover the underlying symmetry group under the assumption that it is orthogonal. By utilizing the corresponding unique symmetric matrix and its inherent diagonal form, we incorporate suitable inductive biases into the neural network architecture, leading to models that are both simplified and efficient. Our approach results in an invariant model that preserves norms, while the equivariant model is represented as a product of a norm-invariant model and a scale-invariant model, where the ``product'' refers to the group action. Moreover, we extend our framework to a more general setting where the function acts on tuples of input vectors via a diagonal (or product) group action. In this extension, the equivariant function is decomposed into an angular component extracted solely from the normalized first vector and a scale-invariant component that depends on the full Gram matrix of the tuple. This decomposition captures the inter-dependencies between multiple inputs while preserving the underlying group symmetry. We assess the effectiveness of our framework across multiple tasks, including polynomial regression, top quark tagging, and moment of inertia matrix prediction. Comparative analysis with baseline methods demonstrates that our model consistently excels in both discovering the underlying symmetry and efficiently learning the corresponding equivariant function.

cs.LG↗

arXiv2025

Latent Mamba Operator for Partial Differential Equations

Neural operators have emerged as powerful data-driven frameworks for solving Partial Differential Equations (PDEs), offering significant speedups over numerical methods. However, existing neural operators struggle with scalability in high-dimensional spaces, incur high computational costs, and face challenges in capturing continuous and long-range dependencies in PDE dynamics. To address these limitations, we introduce the Latent Mamba Operator (LaMO), which integrates the efficiency of state-space models (SSMs) in latent space with the expressive power of kernel integral formulations in neural operators. We also establish a theoretical connection between state-space models (SSMs) and the kernel integral of neural operators. Extensive experiments across diverse PDE benchmarks on regular grids, structured meshes, and point clouds covering solid and fluid physics datasets, LaMOs achieve consistent state-of-the-art (SOTA) performance, with a 32.3% improvement over existing baselines in solution operator approximation, highlighting its efficacy in modeling complex PDE solutions.

cs.LG↗

arXiv2024

UnDIVE: Generalized Underwater Video Enhancement Using Generative Priors

With the rise of marine exploration, underwater imaging has gained significant attention as a research topic. Underwater video enhancement has become crucial for real-time computer vision tasks in marine exploration. However, most existing methods focus on enhancing individual frames and neglect video temporal dynamics, leading to visually poor enhancements. Furthermore, the lack of ground-truth references limits the use of abundant available underwater video data in many applications. To address these issues, we propose a two-stage framework for enhancing underwater videos. The first stage uses a denoising diffusion probabilistic model to learn a generative prior from unlabeled data, capturing robust and descriptive feature representations. In the second stage, this prior is incorporated into a physics-based image formulation for spatial enhancement, while also enforcing temporal consistency between video frames. Our method enables real-time and computationally-efficient processing of high-resolution underwater videos at lower resolutions, and offers efficient enhancement in the presence of diverse water-types. Extensive experiments on four datasets show that our approach generalizes well and outperforms existing enhancement methods. Our code is available at github.com/suhas-srinath/undive.

eess.IV↗

arXiv2024

HOLMES: Hyper-Relational Knowledge Graphs for Multi-hop Question Answering using LLMs

Given unstructured text, Large Language Models (LLMs) are adept at answering simple (single-hop) questions. However, as the complexity of the questions increase, the performance of LLMs degrade. We believe this is due to the overhead associated with understanding the complex question followed by filtering and aggregating unstructured information in the raw text. Recent methods try to reduce this burden by integrating structured knowledge triples into the raw text, aiming to provide a structured overview that simplifies information processing. However, this simplistic approach is query-agnostic and the extracted facts are ambiguous as they lack context. To address these drawbacks and to enable LLMs to answer complex (multi-hop) questions with ease, we propose to use a knowledge graph (KG) that is context-aware and is distilled to contain query-relevant information. The use of our compressed distilled KG as input to the LLM results in our method utilizing up to $67\%$ fewer tokens to represent the query relevant information present in the supporting documents, compared to the state-of-the-art (SoTA) method. Our experiments show consistent improvements over the SoTA across several metrics (EM, F1, BERTScore, and Human Eval) on two popular benchmark datasets (HotpotQA and MuSiQue).

cs.CL↗

arXiv2024

CoroNetGAN: Controlled Pruning of GANs via Hypernetworks

Generative Adversarial Networks (GANs) have proven to exhibit remarkable performance and are widely used across many generative computer vision applications. However, the unprecedented demand for the deployment of GANs on resource-constrained edge devices still poses a challenge due to huge number of parameters involved in the generation process. This has led to focused attention on the area of compressing GANs. Most of the existing works use knowledge distillation with the overhead of teacher dependency. Moreover, there is no ability to control the degree of compression in these methods. Hence, we propose CoroNet-GAN for compressing GAN using the combined strength of differentiable pruning method via hypernetworks. The proposed method provides the advantage of performing controllable compression while training along with reducing training time by a substantial factor. Experiments have been done on various conditional GAN architectures (Pix2Pix and CycleGAN) to signify the effectiveness of our approach on multiple benchmark datasets such as Edges-to-Shoes, Horse-to-Zebra and Summer-to-Winter. The results obtained illustrate that our approach succeeds to outperform the baselines on Zebra-to-Horse and Summer-to-Winter achieving the best FID score of 32.3 and 72.3 respectively, yielding high-fidelity images across all the datasets. Additionally, our approach also outperforms the state-of-the-art methods in achieving better inference time on various smart-phone chipsets and data-types making it a feasible solution for deployment on edge devices.

cs.CV↗

arXiv2023

CoNO: Complex Neural Operator for Continuous Dynamical Systems

Neural operators extend data-driven models to map between infinite-dimensional functional spaces. These models have successfully solved continuous dynamical systems represented by differential equations, viz weather forecasting, fluid flow, or solid mechanics. However, the existing operators still rely on real space, thereby losing rich representations potentially captured in the complex space by functional transforms. In this paper, we introduce a Complex Neural Operator (CoNO), that parameterizes the integral kernel in the complex fractional Fourier domain. Additionally, the model employing a complex-valued neural network along with aliasing-free activation functions preserves the complex values and complex algebraic properties, thereby enabling improved representation, robustness to noise, and generalization. We show that the model effectively captures the underlying partial differential equation with a single complex fractional Fourier transform. We perform an extensive empirical evaluation of CoNO on several datasets and additional tasks such as zero-shot super-resolution, evaluation of out-of-distribution data, data efficiency, and robustness to noise. CoNO exhibits comparable or superior performance to all the state-of-the-art models in these tasks. Altogether, CoNO presents a robust and superior model for modeling continuous dynamical systems, providing a fillip to scientific machine learning.

cs.LG↗

arXiv2023

CoDBench: A Critical Evaluation of Data-driven Models for Continuous Dynamical Systems

Continuous dynamical systems, characterized by differential equations, are ubiquitously used to model several important problems: plasma dynamics, flow through porous media, weather forecasting, and epidemic dynamics. Recently, a wide range of data-driven models has been used successfully to model these systems. However, in contrast to established fields like computer vision, limited studies are available analyzing the strengths and potential applications of different classes of these models that could steer decision-making in scientific machine learning. Here, we introduce CodBench, an exhaustive benchmarking suite comprising 11 state-of-the-art data-driven models for solving differential equations. Specifically, we comprehensively evaluate 4 distinct categories of models, viz., feed forward neural networks, deep operator regression models, frequency-based neural operators, and transformer architectures against 8 widely applicable benchmark datasets encompassing challenges from fluid and solid mechanics. We conduct extensive experiments, assessing the operators' capabilities in learning, zero-shot super-resolution, data efficiency, robustness to noise, and computational efficiency. Interestingly, our findings highlight that current operators struggle with the newer mechanics datasets, motivating the need for more robust neural operators. All the datasets and codes will be shared in an easy-to-use fashion for the scientific community. We hope this resource will be an impetus for accelerated progress and exploration in modeling dynamical systems.

cs.LG↗

arXiv2023

Neural Discovery of Permutation Subgroups

We consider the problem of discovering subgroup $H$ of permutation group $S_{n}$. Unlike the traditional $H$-invariant networks wherein $H$ is assumed to be known, we present a method to discover the underlying subgroup, given that it satisfies certain conditions. Our results show that one could discover any subgroup of type $S_{k} (k \leq n)$ by learning an $S_{n}$-invariant function and a linear transformation. We also prove similar results for cyclic and dihedral subgroups. Finally, we provide a general theorem that can be extended to discover other subgroups of $S_{n}$. We also demonstrate the applicability of our results through numerical experiments on image-digit sum and symmetric polynomial regression tasks.

cs.LG↗

arXiv2021

Unsupervised Domain Adaptation Schemes for Building ASR in Low-resource Languages

Building an automatic speech recognition (ASR) system from scratch requires a large amount of annotated speech data, which is difficult to collect in many languages. However, there are cases where the low-resource language shares a common acoustic space with a high-resource language having enough annotated data to build an ASR. In such cases, we show that the domain-independent acoustic models learned from the high-resource language through unsupervised domain adaptation (UDA) schemes can enhance the performance of the ASR in the low-resource language. We use the specific example of Hindi in the source domain and Sanskrit in the target domain. We explore two architectures: i) domain adversarial training using gradient reversal layer (GRL) and ii) domain separation networks (DSN). The GRL and DSN architectures give absolute improvements of 6.71% and 7.32%, respectively, in word error rate over the baseline deep neural network model when trained on just 5.5 hours of data in the target domain. We also show that choosing a proper language (Telugu) in the source domain can bring further improvement. The results suggest that UDA schemes can be helpful in the development of ASR systems for low-resource languages, mitigating the hassle of collecting large amounts of annotated speech data.

cs.CL↗