SearcharxivSearch

arXiv subjects

Yogesh Kumar

Publications and source records attributed to Yogesh Kumar.

At least 37 records · Page 2Linked to original sources

Language-Guided Temporal Token Pruning for Efficient VideoLLM Processing

Vision Language Models (VLMs) struggle with long-form videos due to the quadratic complexity of attention mechanisms. We propose Language-Guided Temporal Token Pruning (LGTTP), which leverages temporal cues from queries to adaptively prune video tokens, preserving contextual continuity while reducing computational overhead. Unlike uniform pruning or keyframe selection, LGTTP retains higher token density in temporally relevant segments. Our model-agnostic framework integrates with TimeChat and LLaVA-Video, achieving a 65% reduction in computation while preserving 97-99% of the original performance. On QVHighlights, LGTTP improves HIT@1 by +9.5%, and on Charades-STA, it retains 99.6% of R@1. It excels on queries with explicit temporal markers and remains effective across general video understanding tasks.

cs.CV

Stranski-Krastanov Growth of Disordered ScNx Thin Films on MgO(100): Influence of Defect Densities on Electronic Structure and Transport Properties

We report a nascent real time Stranski-Krastanov growth of reactively sputtered ScNx thin films on MgO(100). The epitaxial growth was limited to 5 nm at a substrate temperature (Ts) of 25 C while the self-sustaining epitaxial nature along the [100] azimuth was retained up to 25 nm in Ts = 250 and 500 C samples due to enhanced adatom mobility. At Ts = 700 C, the film showed half order in-situ RHEED pattern, with forbidden (hkl) planes indicating N deficient hcp Sc-N phase. Presence of defect densities i.e., N vacancies and O interstitials leads to a disorder in ScNx system with weak localization effect and appearance of Raman relaxed first order transverse and longitudinal optical phonon modes and further leads to metal like Seebeck coefficient. Higher grain boundaries at Ts = 25 C and higher N out-diffusion at Ts = 700 C paves way for incorporation of higher oxygen interstitial in these samples.

cond-mat.mtrl-sci

VideoLLM Benchmarks and Evaluation: A Survey

The rapid development of Large Language Models (LLMs) has catalyzed significant advancements in video understanding technologies. This survey provides a comprehensive analysis of benchmarks and evaluation methodologies specifically designed or used for Video Large Language Models (VideoLLMs). We examine the current landscape of video understanding benchmarks, discussing their characteristics, evaluation protocols, and limitations. The paper analyzes various evaluation methodologies, including closed-set, open-set, and specialized evaluations for temporal and spatiotemporal understanding tasks. We highlight the performance trends of state-of-the-art VideoLLMs across these benchmarks and identify key challenges in current evaluation frameworks. Additionally, we propose future research directions to enhance benchmark design, evaluation metrics, and protocols, including the need for more diverse, multimodal, and interpretability-focused benchmarks. This survey aims to equip researchers with a structured understanding of how to effectively evaluate VideoLLMs and identify promising avenues for advancing the field of video understanding with large language models.

cs.CV

Design, Contact Modeling, and Collision-inclusive Planning of a Dual-stiffness Aerial RoboT (DART)

Collision-resilient quadrotors have gained significant attention given their potential for operating in cluttered environments and leveraging impacts to perform agile maneuvers. However, existing designs are typically single-mode: either safeguarded by propeller guards that prevent deformation or deformable but lacking rigidity, which is crucial for stable flight in open environments. This paper introduces DART, a Dual-stiffness Aerial RoboT, that adapts its post-collision response by either engaging a locking mechanism for a rigid mode or disengaging it for a flexible mode, respectively. Comprehensive characterization tests highlight the significant difference in post collision responses between its rigid and flexible modes, with the rigid mode offering seven times higher stiffness compared to the flexible mode. To understand and harness the collision dynamics, we propose a novel collision response prediction model based on the linear complementarity system theory. We demonstrate the accuracy of predicting collision forces for both the rigid and flexible modes of DART. Experimental results confirm the accuracy of the model and underscore its potential to advance collision-inclusive trajectory planning in aerial robotics.

cs.RO

Towards Making Flowchart Images Machine Interpretable

Computer programming textbooks and software documentations often contain flowcharts to illustrate the flow of an algorithm or procedure. Modern OCR engines often tag these flowcharts as graphics and ignore them in further processing. In this paper, we work towards making flowchart images machine-interpretable by converting them to executable Python codes. To this end, inspired by the recent success in natural language to code generation literature, we present a novel transformer-based framework, namely FloCo-T5. Our model is well-suited for this task,as it can effectively learn semantics, structure, and patterns of programming languages, which it leverages to generate syntactically correct code. We also used a task-specific pre-training objective to pre-train FloCo-T5 using a large number of logic-preserving augmented code samples. Further, to perform a rigorous study of this problem, we introduce theFloCo dataset that contains 11,884 flowchart images and their corresponding Python codes. Our experiments show promising results, and FloCo-T5 clearly outperforms related competitive baselines on code generation metrics. We make our dataset and implementation publicly available.

cs.CV

A New Algorithm for Computing Branch Number of Non-Singular Matrices over Finite Fields

The notion of branch numbers of a linear transformation is crucial for both linear and differential cryptanalysis. The number of non-zero elements in a state difference or linear mask directly correlates with the active S-Boxes. The differential or linear branch number indicates the minimum number of active S-Boxes in two consecutive rounds of an SPN cipher, specifically for differential or linear cryptanalysis, respectively. This paper presents a new algorithm for computing the branch number of non-singular matrices over finite fields. The algorithm is based on the existing classical method but demonstrates improved computational complexity compared to its predecessor. We conduct a comparative study of the proposed algorithm and the classical approach, providing an analytical estimation of the algorithm's complexity. Our analysis reveals that the computational complexity of our algorithm is the square root of that of the classical approach.

cs.CR

Construction of all MDS and involutory MDS matrices

In this paper, we propose two algorithms for a hybrid construction of all $n\times n$ MDS and involutory MDS matrices over a finite field $\mathbb{F}_{p^m}$, respectively. The proposed algorithms effectively narrow down the search space to identify $(n-1) \times (n-1)$ MDS matrices, facilitating the generation of all $n \times n$ MDS and involutory MDS matrices over $\mathbb{F}_{p^m}$. To the best of our knowledge, existing literature lacks methods for generating all $n\times n$ MDS and involutory MDS matrices over $\mathbb{F}_{p^m}$. In our approach, we introduce a representative matrix form for generating all $n\times n$ MDS and involutory MDS matrices over $\mathbb{F}_{p^m}$. The determination of these representative MDS matrices involves searching through all $(n-1)\times (n-1)$ MDS matrices over $\mathbb{F}_{p^m}$. Our contributions extend to proving that the count of all $3\times 3$ MDS matrices over $\mathbb{F}_{2^m}$ is precisely $(2^m-1)^5(2^m-2)(2^m-3)(2^{2m}-9\cdot 2^m+21)$. Furthermore, we explicitly provide the count of all $4\times 4$ MDS and involutory MDS matrices over $\mathbb{F}_{2^m}$ for $m=2, 3, 4$.

cs.CR

Improving Medical Multi-modal Contrastive Learning with Expert Annotations

We introduce eCLIP, an enhanced version of the CLIP model that integrates expert annotations in the form of radiologist eye-gaze heatmaps. It tackles key challenges in contrastive multi-modal medical imaging analysis, notably data scarcity and the "modality gap" -- a significant disparity between image and text embeddings that diminishes the quality of representations and hampers cross-modal interoperability. eCLIP integrates a heatmap processor and leverages mixup augmentation to efficiently utilize the scarce expert annotations, thus boosting the model's learning effectiveness. eCLIP is designed to be generally applicable to any variant of CLIP without requiring any modifications of the core architecture. Through detailed evaluations across several tasks, including zero-shot inference, linear probing, cross-modal retrieval, and Retrieval Augmented Generation (RAG) of radiology reports using a frozen Large Language Model, eCLIP showcases consistent improvements in embedding quality. The outcomes reveal enhanced alignment and uniformity, affirming eCLIP's capability to harness high-quality annotations for enriched multi-modal analysis in the medical imaging domain.

cs.CV

Investigation of Angle dependent SdH oscillations in Topological Insulator Bismuth

The current article investigated the band structure in the presence and absence of spin-orbit coupling (SOC), examined the Z2 invariants, and investigated the detailed angle-dependent magneto-transport of up to 10 T (Tesla) and down to 2 K for the Bismuth crystal. The out-of-plane field-dependent magnetoresistance (MR) is positive and is huge to the order of ~104% at 2 K and 10 T. On the other hand, the longitudinal (in-plane) field-dependent MR is relatively small and is negative. The thermal activation energy is also estimated by using the Boltzmann formula from resistivity vs temperature measurement under applied transverse magnetic fields. The topological nature of Bi is confirmed by Z2 invariant calculation using Density functional theory. PBESol bands show trivial but Hybrid functional (HSE) bands show non-trivial topology being present in Bismuth. This article comprehensively studies the dependence of MR oscillations upon the angle between the applied field and the current. The observed oscillations fade away as the angle is increased. This article is an extension of our previous work on Bismuth [1], in which we conducted a comprehensive analysis of its structural and micro-structural properties along with its transport behavior in an applied transverse magnetic field.

cond-mat.mtrl-sci

Superconductivity at 9 K in Pb-Bi Alloy

In the present work, we report the synthesis of Pb-Bi alloy with enhanced Tc of up to 9K, which is higher than that of Pb. The alloy is synthesized via a solid-state reaction route in the vacuum-encapsulated quartz tube at 7000C in an automated furnace. The synthesized sample is characterized by X-ray Diffraction(XRD) and Energy dispersive X-ray analysis(EDAX) for its phase purity and elemental composition. Rietveld refinement of XRD reveals that the end product is a majority hexagonal Pb7Bi3, with minor rhombohedral Bi. The electronic transport measurement shows metallic behavior with the Debye temperature of 108K and a superconductivity transition temperature (Tc) below 9K, which is the maximum to date for any reported Pb-Bi alloy, Pb or Bi at ambient pressure. Partial substitution of Bi at the Pb site may modify the free density of electronic states within the BCS model to attain the optimum Tc, which is higher by around 2K from the reported Tc of Pb. The superconductor phase diagram derived from magneto-transport measurements reveals that the synthesized alloy is a conventional superconductor with an upper critical field (Hc2) of 3.9 Tesla, which lies well within the Pauli paramagnetic limit. The magnetization measurements carried out following ZFC(Zero Field Cool) protocols infer that the synthesized alloy is a bulk superconductor below 9K. The isothermal M-H(Magnetization vs. Field) measurements performed below Tc establish it as a type-II superconductor. The specific heat capacity measurements show that the Pb-Bi alloy is a strongly coupled bulk superconductor below around 9K with possibly two superconducting gaps.

cond-mat.supr-con

A Systematic Construction Approach for All $4\times 4$ Involutory MDS Matrices

Maximum distance separable (MDS) matrices play a crucial role not only in coding theory but also in the design of block ciphers and hash functions. Of particular interest are involutory MDS matrices, which facilitate the use of a single circuit for both encryption and decryption in hardware implementations. In this article, we present several characterizations of involutory MDS matrices of even order. Additionally, we introduce a new matrix form for obtaining all involutory MDS matrices of even order and compare it with other matrix forms available in the literature. We then propose a technique to systematically construct all $4 \times 4$ involutory MDS matrices over a finite field $\mathbb{F}_{2^m}$. This method significantly reduces the search space by focusing on involutory MDS class representative matrices, leading to the generation of all such matrices within a substantially smaller set compared to considering all $4 \times 4$ involutory matrices. Specifically, our approach involves searching for these representative matrices within a set of cardinality $(2^m-1)^5$. Through this method, we provide an explicit enumeration of the total number of $4 \times 4$ involutory MDS matrices over $\mathbb{F}_{2^m}$ for $m=3,4,\ldots,8$.

cs.CR

Role of time-varying magnetic field on QGP equation of state

The phase diagram of quantum chromodynamics (QCD) and its associated thermodynamic properties of quark gluon plasma (QGP) are studied in the presence of time dependent magnetic field. The study plays a pivotal role in the field of cosmology, astrophysics, and heavy ion collisions. In order to explore the structure of quark gluon plasma to deal with the dynamics of quarks and gluons, we investigate the equation of state (EoS) not only in the environment of static magnetic field but also in the presence of time-varying magnetic fields. So, for determining the equation of state of QGP at non zero magnetic fields, we revisited our earlier model where the effect of time varying magnetic field was not taken into consideration. Using the phenomenological model, some appealing features are noticed depending upon the three different scales; effective mass of quark, temperature, and time independent as well as time-dependent magnetic field. Earlier the effective mass of quark was incorporated in our calculations and in the current work, it is modified for static and time-varying magnetic fields. Thermodynamic observables including pressure, energy density, entropy, etc. are calculated for a wide range of temperature and time-dependent as well as time-independent magnetic fields. Finally, we claim that the EoS are highly affected in the presence of a magnetic field. Our results are notable compared to other approaches and found to be advantageous for the measurement of QGP equation of state. These crucial findings with and without time-varying magnetic field could have phenomenological implications in various sectors of high energy physics.

hep-ph

Weak Anti-Localization Effect in Topological Ni$_3$In$_2$S$_2$ Single Crystal

Ni$_3$In$_2$S$_2$ is the most recent entrant into the family of topological insulator (TI) materials, the same exhibits very high MR in a low-temperature regime. Here, we report the crystal growth, the structural, micro-structural, and magneto-transport study of Ni$_3$In$_2$S$_2$ down to 2.5K under an applied field of up to 14Tesla. The phase purity and growth direction of a single crystal is studied by performing XRD on both powder and flake and further Rietveld analysis is also carried out. The electrical transport measurements are studied and the grown crystal showed metallic behaviour down to 2.5K, with an R300K/R2K ratio of around 7. A significant variation in magnetoresistance (MR) values is observed as the temperature is increased from 2.5K to 200K under an applied field of up to 14 Tesla. Interestingly the low T (2.5K), MR shows a clear V-type characteristic TI cusp. Magnetoconductivity data at low fields (1Tesla) is fitted with the Hikami Larkin Nagaoka (HLN) model, which showed the presence of a weak anti-localization effect in the synthesized Ni$_3$In$_2$S$_2$ crystal at low temperatures. We have successfully grown near single-phase Ni$_3$In$_2$S$_2$ and its TI behavior is demonstrated by magneto-transport measurements.

cond-mat.mes-hall

Geometric Tracking Control of a Multi-rotor UAV for Partially Known Trajectories

This paper presents a trajectory-tracking controller for multi-rotor unmanned aerial vehicles (UAVs) in scenarios where only the desired position and heading are known without the higher-order derivatives. The proposed solution modifies the state-of-the-art geometric controller, effectively addressing challenges related to the non-existence of the desired attitude and ensuring positive total thrust input for all time. We tackle the additional challenge of the non-availability of the higher derivatives of the trajectory by introducing novel nonlinear filter structures. We formalize theoretically the effect of these filter structures on the system error dynamics. Subsequently, through a rigorous theoretical analysis, we demonstrate that the proposed controller leads to uniformly ultimately bounded system error dynamics.

eess.SY

SANSformers: Self-Supervised Forecasting in Electronic Health Records with Attention-Free Models

Despite the proven effectiveness of Transformer neural networks across multiple domains, their performance with Electronic Health Records (EHR) can be nuanced. The unique, multidimensional sequential nature of EHR data can sometimes make even simple linear models with carefully engineered features more competitive. Thus, the advantages of Transformers, such as efficient transfer learning and improved scalability are not always fully exploited in EHR applications. Addressing these challenges, we introduce SANSformer, an attention-free sequential model designed with specific inductive biases to cater for the unique characteristics of EHR data. In this work, we aim to forecast the demand for healthcare services, by predicting the number of patient visits to healthcare facilities. The challenge amplifies when dealing with divergent patient subgroups, like those with rare diseases, which are characterized by unique health trajectories and are typically smaller in size. To address this, we employ a self-supervised pretraining strategy, Generative Summary Pretraining (GSP), which predicts future summary statistics based on past health records of a patient. Our models are pretrained on a health registry of nearly one million patients, then fine-tuned for specific subgroup prediction tasks, showcasing the potential to handle the multifaceted nature of EHR data. In evaluation, SANSformer consistently surpasses robust EHR baselines, with our GSP pretraining method notably amplifying model performance, particularly within smaller patient subgroups. Our results illuminate the promising potential of tailored attention-free models and self-supervised pretraining in refining healthcare utilization predictions across various patient demographics.

cs.LG

Spin wave behavior of a novel hopfion-like chiral state in Co/Pt nanodiscs

This work discusses the rich phase diagram of non-trivial chiral spin textures in confined ferromagnetic/heavy-metal (FM/HM) bilayer nanomagnets of circular cross-section. These spin textures are realized as a minimum-energy ground state during an external bias field sweep for a range of nanomagnet's diameter (d). Our study, based on micromagnetic simulations, has revealed a novel Hopfion-like state which can be stabilized for a wide range of diameters and external magnetic fields. We explored the dynamical characteristics of this novel Hopfion-like state under a transient magnetic field applied along the plane's perpendicular direction. Simulation results have demonstrated the excitation of nonreciprocal spin wave (SW) modes for this novel chiral state, in contrast to other stabilized chiral states. These modes are characterized as breathing and quantized radial modes, which also exhibit hybridization with azimuthal modes. The resonant SW modes have been used to demonstrate the switching from a Hopfion-like state to a skyrmion within a few nanoseconds of SW excitation. Furthermore, we establish a correlation between the behavior of excited SW modes as a function of external magnetic field strength and underlying chiral spin texture states.

cond-mat.mes-hall

Emerging interfacial magnetization in isovalent manganite heterostructures driven by octahedral coupling

The distortion of corner-sharing octahedra in isovalent perovskite transition-metal oxide interfaces is proven to be an excellent way to tailor the electronic and magnetic properties of their heterostructures. Combining depth-dependent magnetic characterization technique; (polarized neutron reflectivity, PNR); and theoretical calculation (density functional theory), we report interface-driven magnetic exchange interactions due to a modification in the octahedral rotations at the interfaces in an isovalent La0.67Ca0.33MnO3 (LCMO)/La0.67Sr0.33MnO3 (LSMO) heterostructures. PNR results determined a length scale of ~ 8 unit cells at the interface, which demonstrated a modification in magnetic properties. The results also predicted a low-temperature exchange bias for these ferromagnetic heterostructures with a maximum exchange bias for the heterostructure, which showed an enhanced antiferromagnetic coupling at the interfaces.

cond-mat.mtrl-sci

Absence of superconductivity in LK-99 at ambient conditions

The report of synthesis of modified Lead apatite (LK-99) with evidence of superconductivity at more than boiling water temperature has steered the whole scientific community. There have been several failures to reproduce superconductivity in LK-99 including partial successes. Here, we have continued our efforts to synthesize phase pure LK-99 with improved precursors. The process has been followed as suggested by Sukbae Lee et. al., [1,2]. The phase purity of each precursor is evidenced by Powder X-ray diffraction (PXRD) and well fitted by Rietveld refinement. The PXRD confirms the synthesis of phase pure polycrystalline LK-99 with apatite structure. The freshly synthesized sample does not show any signature of superconductivity levitation on a magnet (diamagnetism). The magnetization measurements on SQUID also show that LK-99 is diamagnetic at 280 K, there is no sign of superconductivity in LK-99 at room temperature. Moreover, we have also performed first principle calculations to investigate the electronic band structure of the LK-99 near Fermi level. Our study verifies that the Cu doped lead apatite (LK-99) has bands crossing at Fermi level, indicating generation of strong correlation in the system.

cond-mat.supr-con