Abstract

With over 1.4 billion people and dozens of widely used scripts, India represents one of the world's most linguistically and visually diverse document ecosystems. Yet document understanding benchmarks remain largely Western- and English-centric. Consequently, the robustness of modern vision-language models (VLMs) in multilingual and low-resource settings remains under-evaluated. We introduce the first large-scale multilingual benchmark for Indian document understanding, comprising 52607 documents across 23 languages, 13 domains, 71 document classes, and 52 tasks, with over 2.4M QA annotations across multiple tasks. The dataset captures diverse layouts, scripts, and visual conditions spanning governance, education, finance, archival, and informal content, among others. We additionally include synthetic multilingual charts, tables, and diagrams generated via Patram-Syn, an Indian persona-driven data generation pipeline with human validation. The benchmark measures perception-level capabilities such as OCR, parsing, key information extraction, and layout detection, as well as reasoning-level capabilities including Document VQA with abstractive, multi-hop, and ambiguous questions. We further introduce targeted settings for cross-lingual, transliterated, and code-mixed VQA. We also propose two new evaluation metrics, Document Understanding Cross-Lingual Index (DUCLI) and Script Robustness Score (SRS), to quantify cross-lingual reasoning degradation and script robustness. Our benchmark exposes consistent weaknesses in both open-source and closed-source VLMs, particularly in low-resource languages, degraded scans, and layout-intensive settings, underscoring fundamental challenges in scalable multilingual document intelligence.

Abstract

System identification for Flapping Wing Unmanned
Aerial Vehicles (FWUAVs) typically relies on simplified unsteady
aerodynamics into time-averaged models. We propose an aerodynamic framework that identifies unsteady force parameters,
including apparent mass and aeroelastic effects, using IMU and
airspeed data. Utilizing a modified DeLaurier's formulation,
our model estimates aerodynamic coefficients that approximate
unsteady effects within a computationally efficient framework.
Model parameters are optimized using the Covariance Matrix
Adaptation Evolution Strategy (CMA-ES). The framework was
trained and validated using experimental flight data from a 1.4
m wingspan prototype.The results demonstrate good predictive
capability, achieving a reduction in Root Mean Square Error
(RMSE) of 36% when compared to the baseline model without
incorporating unsteady effects

Abstract

An analytical framework for seismic rocking analysis of nonstructural elements in the monuments
of the archaeological site of Bagan, identified as vulnerable in past earthquakes, is presented. While
the overarching study focuses on four structural typologies, a nonlinear kinematic analysis incorporating a finite compressive-strength rocking model is adopted specifically for the susceptible
rocking elements of three selected representative monuments. Seismic demand is evaluated using
three definitions: the recorded 2016 Chauk earthquake ground motion, monument-specific simulated records, and the design spectrum based on Myanmar's National Building Code of 2020
(MNBC 2020). The floor spectra are computed for each rocking element using the procedure
based on Italian norms for construction (CNTC 2019) accounting for height-dependent acceleration amplification. Results show that the main spires of all three monuments are the most
vulnerable elements, with amplification factors of 1.65-2.45 relative to ground level.
A comparison of the three demand definitions reveals distinct outcomes: The design spectrum
definition consistently yields the most conservative demand estimate due to its higher peak
ground acceleration, while the spatial variability captured by monument-specific simulated records
reveals significant inter-monument differences not apparent from the single record of Chauk
earthquake

Abstract

This review paper discusses the evolution of Integrated Modular Avionics (IMA) architecture for aerospace applications and the adoption of IMA in missile applications. It provides an overview of the traditional Federated Architecture and also explains the concept of Integrated Modular Avionics (IMA) architecture. The paper also discusses about the design, implementation, methodology, and compares the pros and cons of both architectures. It also highlights the characteristics of IMA, the design challenges, and how these challenges are overcome by providing solutions. The methods adopted in building IMA systems and the improvements made are clearly explained. This review paper presents the results of a pilot implementation and provides recommendations for future adoption of this technology in missile applications. The paper also discusses about the lessons learned from various commercial IMA systems (worldwide), these learnings can be adopted for successful IMA system development for military applications.

Abstract

Monte Carlo (MC) sampling and related enhanced sampling methods can efficiently
explore conformational space and yield thermodynamically accurate populations, but
they produce no temporal information and therefore cannot directly characterize molec-
ular kinetics. Here we present a framework for constructing a continuous-time Markov
chain (CTMC) from a static, Boltzmann-weighted ensemble, enabling the generation of
physically plausible kinetic trajectories without requiring extensive molecular dynamics
(MD) simulation. The method includes (i) clustering the MC ensemble into metastable
discrete states whose populations define the stationary distribution, (ii) constructing a
sparse state-transition graph from the intersection of Delaunay-geometric and empirically
observed adjacencies, (iii) building a rate matrix over this graph that satisfies detailed
balance by design, with transition rates modulated by a proximity kernel over the
conformational landscape, and (iv) calibrating the absolute timescale by matching the
slowest relaxation mode to a short MD reference trajectory. We validate the approach
on two test systems, n-pentane and alanine dipeptide, demonstrating quantitative agree-
ment with MD-derived rate matrices, state occupancy autocorrelation functions, and implied timescales. The method provides a principled route to kinetic insight when MD
convergence is prohibitively expensive but thermodynamic sampling is tractable, and
serves as a complement to existing Markov state model approaches that build kinetic
models from MD data.

Abstract

Developments in Graph-Language Models (GLMs) aim to integrate the structural reasoning capabilities of Graph Neural Networks (GNNs) with the semantic understanding of Large Language Models (LLMs). However, we demonstrate that current evaluation benchmarks for GLMs, which are primarily repurposed node-level classification datasets, are insufficient to assess multimodal reasoning. Our analysis reveals that strong performance on these benchmarks is achievable using unimodal information alone, suggesting that they do not necessitate graph-language integration. To address this evaluation gap, we introduce the CLEGR(Compositional Language-Graph Reasoning) benchmark, designed to evaluate multimodal reasoning at various complexity levels. Our benchmark employs a synthetic graph generation pipeline paired with questions that require joint reasoning over structure and textual semantics. We perform a thorough evaluation of representative GLM architectures and find that soft-prompted LLM baselines perform on par with GLMs that incorporate a full GNN backbone. This result calls into question the architectural necessity of incorporating graph structure into LLMs. We further show that GLMs exhibit significant performance degradation in tasks that require structural reasoning. These findings highlight limitations in the graph reasoning capabilities of current GLMs and provide a foundation for advancing the community toward explicit multimodal reasoning involving graph structure and language.

Abstract

Visualising all dimensions of multi-dimensional real data while providing semantics is a challenging problem in data analysis. In this paper, we present the visualisation of multi-dimensional real data using our tool, 360◦ RAYS. RAYS plots all features and points of the dataset and provides information such as the geometric positioning of points or classes in the orthants of space, thereby aiding understanding of the closeness of points, classification of points, and ordering of subspaces. This visualisation technique is built after exploring viable options to plot multi-dimensional data, which included parallel coordinates, concentric coordinates, and circle segments, among others. The key idea of our approach is a hierarchical concentric ring structure, where each successive ring incrementally adds a dimension to the existing subspace, ordered by the increasing or decreasing coefficient of variation of data in the added dimension. Normalised data points are mapped to ring sectors based on a bitwise encoding of their coordinate values (non-negative or negative). The advantage of RAYS is that it presents the complete data with all dimensions and provides a clear visual grouping of points by subspace concentration. The visualisation also shows pure and overlapping orthants within the data in a flattened 2D radial layout.

Abstract

Speech enhancement models optimised for percep-
tual quality metrics such as PESQ do not necessarily improve
automatic speech recognition (ASR) performance, as aggres-
sive noise suppression can distort the phonemic cues that
acoustic models rely on. We propose a time-domain speech
frontend based on the Mamba selective state space model,
trained exclusively with ASR-oriented loss functions to pre-
serve phonetically discriminative structure rather than max-
imise perceptual quality. The model follows a U-Net encoder-
decoder architecture with 4× temporal downsampling and six
stacked bidirectional Mamba blocks operating on the raw wave-
form, trained with a multi-component loss combining multi-
resolution spectral supervision, log-mel feature matching, speech
correlation preservation, and ideal-ratio-mask spectral objec-
tives. On VoiceBank-DEMAND, the proposed model achieves
10.26% WER with wav2vec2-base-960h and 7.68% with
whisper-base, achieving statistically equivalent WER to SE-
Mamba (10.30% / 7.73%) while producing a lower PESQ
of 3.10 versus 3.69. Phoneme-class analysis confirms that the
proposed model matches or outperforms SEMamba on five
of seven articulatory classes, demonstrating that ASR-oriented
training objectives achieve equivalent robustness to perceptual
enhancement models without modifying the downstream ASR
system.

Abstract

Subtitles are essential for accessibility and global dissemination of video content, yet conventional formats often omit critical cues such as speaker identities. Character-Aware Subtitling (CAS) aims to address this gap, however, existing approaches are limited by the absence of public checkpoints, weak pipeline coordination leading to detrimental performance, and insufficient failure analysis. This work presents a systematic study of subtitling pipelines, demonstrating that improved transcript-speech alignment substantially enhances downstream performance, achieving state-of-the-art results on the LLR dataset by 2.77% DER-point zero-shot, and by 4.78% DER-point after VAD finetuning. In addition, we propose design refinements, adopt open, out-of-the-box model alternatives, and introduce a scalable method for curating diverse character priors, thereby improving the robustness and replicability of CAS.

Abstract

We study the connection between audio observations and the underlying physics of a mundane yet intriguing everyday activity: pouring liquids. Given only the sound of liquid pouring into a container, our objective is to automatically infer physical properties such as the liquid level, the shape and size of the container, the pouring rate and the time to fill. To this end, we: (i) show in theory that these properties can be determined from the fundamental frequency (pitch); (ii) train a pitch detection model with supervision from simulated data and visual data with a physics-inspired objective; (iii) introduce a new large dataset of real pouring videos for a systematic study; (iv) show that the trained model can indeed infer these physical properties for real data; and finally, (v) we demonstrate strong generalization to various container shapes, other datasets, and in-the-wild YouTube videos. Our work presents a keen understanding of a narrow yet rich problem at the intersection of acoustics, physics, and learning. It opens up applications to enhance multisensory perception in robotic pouring.