arXiv subjects

Rajeev Ranjan

Publications and source records attributed to Rajeev Ranjan.

At least 19 recordsLinked to original sources

arXiv2026

First principles study of thermoelectric properties of $\text{Nb}_2\text{Co}_2\text{InSb}$ and $\text{Nb}_2\text{Co}_2\text{GaSb}$ double half-Heuslers

Valence electron count (VEC) 18 half-Heusler (hH) alloys are considered promising for high-temperature thermoelectric applications due to their high Seebeck coefficient, mechanical stability, and robustness. However, their relatively large lattice thermal conductivity ($k_{L}$) significantly limits their thermoelectric performance. Introducing mass disorder at lattice sites is an effective approach to suppress $k_{L}$ through enhanced phonon scattering. For instance, NbCoSn exhibits a low figure of merit ($zT \sim 0.05$) despite having a reasonably high power factor of 2.1~$\text{mW}/\text{mK}^{2}$ at room temperature, mainly due to its large lattice thermal conductivity, reported to be 13.25~$\text{W}/\text{mK}$ experimentally and 18~$\text{W}/\text{mK}$ theoretically. In this work, we explore the thermoelectric properties of $\text{Nb}_2\text{Co}_2\text{InSb}$ and $\text{Nb}_2\text{Co}_2\text{GaSb}$, which can be regarded as derivatives of NbCoSn with substitution at the Sn site. Both ordered and Special Quasirandom Structures (SQSs) are considered to understand the role of configurational disorder. Energetic analysis indicates that the ordered phase is most stable for $\text{Nb}_2\text{Co}_2\text{InSb}$, whereas the SQS phase is energetically favored for $\text{Nb}_2\text{Co}_2\text{GaSb}$. The lattice thermal conductivity is evaluated using the Debye-Callaway model, yielding values in the range of 5.5-6.9~$\text{W}/\text{mK}$ for $\text{Nb}_2\text{Co}_2\text{InSb}$ and 4.7-5.8~$\text{W}/\text{mK}$ for $\text{Nb}_2\text{Co}_2\text{GaSb}$ at room temperature. These values are significantly lower than those of the parent NbCoSn system, highlighting the effectiveness of mass disorder in reducing thermal conductivity. The results suggest that these double half-Heusler compounds are promising candidates for improved thermoelectric performance.

cond-mat.mtrl-sci↗

arXiv2025

One Word Is Not Enough: Simple Prompts Improve Word Embeddings

Text embedding models are designed for sentence-level applications like retrieval and semantic similarity, and are primarily evaluated on sentence-level benchmarks. Their behavior on isolated words is less understood. We show that simply prepending semantic prompts to words before embedding substantially improves word similarity correlations. Testing 7 text embedding models, including text-embedding-3-large (OpenAI), embed-english-v3.0 (Cohere), voyage-3(Voyage AI), all-mpnet-base-v2, and Qwen3-Embedding-8B, on 3 standard benchmarks (SimLex-999, WordSim-353, MEN-3000), we find that prompts like "meaning: {word}" or "Represent the semantic concept: {word}" improve Spearman correlations by up to +0.29 on SimLex-999. Some models fail completely on bare words (correlation = 0) but recover with prompts (+0.73 improvement). Our best results achieve correlation = 0.692 on SimLex-999 with embed-english-v3.0 (Cohere), correlation = 0.811 on WordSim-353, and correlation = 0.855 on MEN-3000 with text-embedding-3-large (OpenAI). These results outperform classic static embeddings like Word2Vec (correlation = 0.40) and even the best static method LexVec (correlation = 0.48) on SimLex-999, establishing a new state-of-the-art for pure embedding methods. This zero-shot technique requires no training and works with any text embedding model.

cs.CL↗

arXiv2025

Entropy Stabilized ZrHfCoNiSnSb Half-Heusler Alloy for Thermoelectric Applications: A Theoretical Prediction

Half-Heusler (HH) alloys are potential thermoelectric materials for use at elevated temperatures due to their high Seebeck coefficient and superior mechanical and thermal stability. However, their enhanced lattice thermal conductivity is detrimental to thermoelectric applications. One way to circumvent this problem is to introduce mass disorder at lattice sites by mixing the components of two or more alloys. Such systems are typically stabilized by the entropy of mixing. In this work, using computational tools, we propose a mixed HH, namely, ZrHfCoNiSnSb, which can be formed by the elemental compositions of the parent half-Heuslers ZrNiSn/HfNiSn and HfCoSb/ZrCoSb. We propose that this new compound can be synthesized at elevated temperatures, as its Gibbs free energy is reduced due to higher configurational entropy, making it more thermodynamically stable than the parent compounds under such conditions. Our calculations indicate that it is a dynamically stable semiconductor with a band gap of 0.61 eV. Its lattice thermal conductivity at room temperature is $5.39~\text{Wm}^{-1}\text{K}^{-1}$, which is significantly lower than those of the parent compounds. The peak value of this alloy's figure of merit (ZT) is 1.00 for the n-type carriers at 1100 K, which is 27% more than the best figure of merit obtained for the parent compounds.

cond-mat.mtrl-sci↗

arXiv2024

Data Pruning via Separability, Integrity, and Model Uncertainty-Aware Importance Sampling

This paper improves upon existing data pruning methods for image classification by introducing a novel pruning metric and pruning procedure based on importance sampling. The proposed pruning metric explicitly accounts for data separability, data integrity, and model uncertainty, while the sampling procedure is adaptive to the pruning ratio and considers both intra-class and inter-class separation to further enhance the effectiveness of pruning. Furthermore, the sampling method can readily be applied to other pruning metrics to improve their performance. Overall, the proposed approach scales well to high pruning ratio and generalizes better across different classification models, as demonstrated by experiments on four benchmark datasets, including the fine-grained classification scenario.

cs.CV↗

arXiv2023

Ultrahigh electrostrain in Pb-free piezoceramics: Effect of bending

Recently several reports showing ultra-high electrostrain (> 1 %) have appeared in Pb-free piezoceramics. However, there is lack of clarity on the nature of the ultrahigh strain. Here, we demonsrate that the ultrahigh strain is a consequence of bending of the disc. We show that the propensity for bending arises from the difference in the response magnitude of the grains at the positive and negative surfaces of the piezoceramic when the field is applied.

physics.app-ph↗

arXiv2023

Giant electrostriction in bulk RE (III) substituted CeO2: effect of RE 3+ and its concentration

Recent discovery of giant electrostriction in rare earth (RE (III)) substituted ceria (CeO2) thin films driven by electroactive defect complexes and their coordinated elastic response, expands the material spectrum for electrostrain applications beyond the conventional piezoelectric materials. Especially Gd substituted CeO2, with Gd concentration >10% seems to be an ideal material to obtain such large electrostrain response. However, there are not many experimental studies that systematically investigate the effect of RE (III) ion-defect interaction and RE concentration on electrostriction. Here we perform structure-property correlation studies in bulk ceramics of RE3+ substituted ceria doped with RE=Y, La and Gd at various concentrations upto a maximum of 20%, to understand the features responsible for giant electrostriction. Our results show that Y substituted ceria, with atleast 20% Y substitution, is clearly both a giant M and a Q electrostrictor at low frequencies (<20 Hz), and this correlates with the unique attractive defect-dopant interaction of Y with oxygen vacancies. La has a repulsive interaction with oxygen vacancies, and La doped ceria at all the studied compositions (upto 20%) does not show giant electrostiction. Gd has a neutral interaction, and only 20% Gd doped ceria at best falls at the border of classification between giant and non-giant electrostrictors at frequencies <0.05 Hz. Our work takes a step back from thin-films and assesses the fundamental defect features required in the design of giant electrostrictors.

cond-mat.mtrl-sci↗

arXiv2023

Ultrahigh electrostrain > 1% in lead-free piezoceramics: A critical review

Recently, a series of reports showing ultra-high electrostrain (> 1 %) have appeared in several Pb-free piezoceramics. The ultrahigh electrostrain has been attributed exclusively to the defect dipoles created in these systems. We examine these claims based on another report arXiv:2208.07134 which demonstrated that the measured electric field driven strain increased dramatically simply by reducing the thickness of the ceramic discs. We prepared some representative Pb-free compositions reported to exhibit ultrahigh strain and performed electrostrain measurements. We found that these compositions do not show ultrahigh electrostrain if the thickness of the discs is above 0.3 mm (the disc diameters were in the range 10- 12 mm diameter). The ultrahigh strain values were obtained when the thickness was below 0.3 mm. We compare the electrostrain obtained from specimens designed to exhibit defect dipoles with specimens that were not designed to have defect dipoles in Na0.5Bi0.5TiO3 (NBT) and K0.5Na0.5NbO3 (KNN) -based lead-free systems and could obtain much higher strain levels (4- 5 %) in the defect dipole free piezoceramics in the small thickness regime. Our results do not favor the defect dipole theory as the exclusive factor for causing ultrahigh strain in piezoceramics. A new approach is called for to understand the phenomenon of ultrahigh electrostrain caused by the thickness reduction of piezoceramic discs.

cond-mat.mtrl-sci↗

arXiv2023

Giant electromechanical response from defective non-ferroelectric epitaxial BaTiO3 integrated on Si 100

Lead free, silicon compatible materials showing large electromechanical responses comparable to, or better than conventional relaxor ferroelectrics, are desirable for various nanoelectromechanical devices and applications. Defect-engineered electrostriction has recently been gaining popularity to obtain enhanced electromechanical responses at sub 100 Hz frequencies. Here, we report record values of electrostrictive strain coefficients (M31) at frequencies as large as 5 kHz (1.04 x 10-14 m2 per V2 at 1 kHz, and 3.87 x 10-15 m2 per V2 at 5 kHz) using A-site and oxygen-deficient barium titanate thin-films, epitaxially integrated onto Si. The effect is robust and retained even after cycling the devices >5000 times. Our perovskite films are non-ferroelectric, exhibit a different symmetry compared to stoichiometric BaTiO3 and are characterized by twin boundaries and nano polar-like regions. We show that the dielectric relaxation arising from the defect-induced features correlates very well with the observed giant electrostrictive response. These films show large coefficient of thermal expansion (2.36 x 10-5/K), which along with the giant M31 implies a considerable increase in the lattice anharmonicity induced by the defects. Our work provides a crucial step forward towards formulating guidelines to engineer large electromechanical responses even at higher frequencies in lead-free thin films.

cond-mat.mtrl-sci↗

arXiv2022

Minutiae-Guided Fingerprint Embeddings via Vision Transformers

Minutiae matching has long dominated the field of fingerprint recognition. However, deep networks can be used to extract fixed-length embeddings from fingerprints. To date, the few studies that have explored the use of CNN architectures to extract such embeddings have shown extreme promise. Inspired by these early works, we propose the first use of a Vision Transformer (ViT) to learn a discriminative fixed-length fingerprint embedding. We further demonstrate that by guiding the ViT to focus in on local, minutiae related features, we can boost the recognition performance. Finally, we show that by fusing embeddings learned by CNNs and ViTs we can reach near parity with a commercial state-of-the-art (SOTA) matcher. In particular, we obtain a TAR=94.23% @ FAR=0.1% on the NIST SD 302 public-domain dataset, compared to a SOTA commercial matcher which obtains TAR=96.71% @ FAR=0.1%. Additionally, our fixed-length embeddings can be matched orders of magnitude faster than the commercial system (2.5 million matches/second compared to 50K matches/second). We make our code and models publicly available to encourage further research on this topic: https://github.com/tba.

cs.CV↗

arXiv2022

Ultrahigh measured electrostrain in polycrystalline bulk piezoceramics: Role of mechanically relaxed grains

Recently we found that when conventional bulk polycrystalline piezoceramics discs (~ 10 mm diameter) are thinned down to thickness ~ 200 microns or below, the measured unipolar electrostrain (UES) values increased phenomenally (arXiv:2208.07134). Here we demonstrate that this anomalous phenomenon correlates strongly with the decrease in the degree of mutual mechanical clamping between the grains which in turn enables enhanced switching of the ferroelastic domains.

cond-mat.mtrl-sci↗

arXiv2022

Ultrahigh measured unipolar strain greater than 2 % in polycrystalline bulk piezoceramics: Effects of disc dimension

Bulk polycrystalline piezoelectric ceramics are extensively used in wide ranging applications as actuators, transducers, and sensors. For actuator applications, it is desirable that the piezoelectric ceramic gives as large electric field driven unipolar strain as possible. In recent years there is an increasing interest in the design and development of such piezoceramic materials. Here we show that piezoceramic discs (10 mm diameter) of PZT, BaTiO3-based and Na0.5Bi0.5TiO3-based polycrystalline piezoceramics can show very large unipolar strain 2 to 3 % when the thickness of the discs is reduced to 200 microns or below. This is accompanied by increasing asymmetry in the bipolar strain-field loops. The motivation of this brief report is make the research community aware of this important effect in piezoceramics and encourage discussion/research on the this interesting phenomenon, which is applicable to all piezoceramics.

cond-mat.mtrl-sci↗

arXiv2020

High Pressure Structural Investigation on Lead-Free Piezoelectric $0.5Ba(Ti_{0.8}Zr_{0.2})O_3$-$0.5(Ba_{0.7}Ca_{0.3})TiO_3$

The solid solution $0.5Ba(Ti_{0.8}Zr_{0.2})O_3$-$0.5(Ba_{0.7}Ca_{0.3})TiO_3$ (BCZT) has become a promising member of the lead-free piezoelectric materials because of its exceptionally high piezoelectric properties. In this study, we focus on studying pressure-dependent Raman spectroscopy, powder x-ray diffraction and dielectric constant measurements on BCZT. The data show several structural transitions are present, where the system from ambient mixed phase (tetragonal, {\it {P4mm}}+ orthorhombic {\it {Amm2}}) transforms into single phase ({\it {P4mm}}) at 0.26 GPa, then converts into cubic phase ({\it {Pm3m}}) at 4.7 GPa followed by another possible structural re-ordering around 10 GPa. Although there have been a lot of unanimity with the ambient crystallographic state of BCZT, our analysis justifies the presence of an intermediate orthorhombic phase in the Morphological Phase Boundary (MPB) of BCZT phase diagram. The transformation tetragonal to cubic is indicated by the Raman mode softening, unit cell volume change and the $(Ti/Zr)O_6$ octahedra distortion, which coincides with the well-known ferroelectric-paraelectric transition of the system. The sudden drop in the dielectric constant value at 4.7 GPa also confirms the loss of ferroelectric nature of the BCZT ceramic.

cond-mat.mtrl-sci↗

arXiv2019

An Automatic System for Unconstrained Video-Based Face Recognition

Although deep learning approaches have achieved performance surpassing humans for still image-based face recognition, unconstrained video-based face recognition is still a challenging task due to large volume of data to be processed and intra/inter-video variations on pose, illumination, occlusion, scene, blur, video quality, etc. In this work, we consider challenging scenarios for unconstrained video-based face recognition from multiple-shot videos and surveillance videos with low-quality frames. To handle these problems, we propose a robust and efficient system for unconstrained video-based face recognition, which is composed of modules for face/fiducial detection, face association, and face recognition. First, we use multi-scale single-shot face detectors to efficiently localize faces in videos. The detected faces are then grouped respectively through carefully designed face association methods, especially for multi-shot videos. Finally, the faces are recognized by the proposed face matcher based on an unsupervised subspace learning approach and a subspace-to-subspace similarity metric. Extensive experiments on challenging video datasets, such as Multiple Biometric Grand Challenge (MBGC), Face and Ocular Challenge Series (FOCS), IARPA Janus Surveillance Video Benchmark (IJB-S) for low-quality surveillance videos and IARPA JANUS Benchmark B (IJB-B) for multiple-shot videos, demonstrate that the proposed system can accurately detect and associate faces from unconstrained videos and effectively learn robust and discriminative features for recognition.

cs.CV↗

arXiv2019

Crystal Loss and Quality Pooling for Unconstrained Face Verification and Recognition

In recent years, the performance of face verification and recognition systems based on deep convolutional neural networks (DCNNs) has significantly improved. A typical pipeline for face verification includes training a deep network for subject classification with softmax loss, using the penultimate layer output as the feature descriptor, and generating a cosine similarity score given a pair of face images or videos. The softmax loss function does not optimize the features to have higher similarity score for positive pairs and lower similarity score for negative pairs, which leads to a performance gap. In this paper, we propose a new loss function, called Crystal Loss, that restricts the features to lie on a hypersphere of a fixed radius. The loss can be easily implemented using existing deep learning frameworks. We show that integrating this simple step in the training pipeline significantly improves the performance of face verification and recognition systems. We achieve state-of-the-art performance for face verification and recognition on challenging LFW, IJB-A, IJB-B and IJB-C datasets over a large range of false alarm rates (10-1 to 10-7).

cs.CV↗

arXiv2019

Localization: A Missing Link in the Pipeline of Object Matching and Registration

Image registration is a process of aligning two or more images of same objects using geometric transformation. Most of the existing approaches work on the assumption of location invariance. These approaches require object-centric images to perform matching. Further, in absence of intensity level symmetry between the corresponding points in two images, the learning based registration approaches rely on synthetic deformations, which often fail in real scenarios. To address these issues, a combination of convolutional neural networks (CNNs) to perform the desired registration is developed in this work. The complete objective is divided into three sub-objectives: object localization, segmentation and matching transformation. Object localization step establishes an initial correspondence between the images. A modified version of single shot multi-box detector is used for this purpose. The detected region is cropped to make the images object-centric. Subsequently, the objects are segmented and matched using a spatial transformer network employing thin plate spline deformation. Initial experiments on MNIST and Caltech-101 datasets show that the proposed model is able to produce accurate matching. Quantitative evaluation performed using dice coefficient (DC) and mean intersection over union (mIoU) show that proposed method results in the values of 79% and 66%, respectively for MNIST dataset and the values of 94% and 90%, respectively for Caltech-101 dataset. The proposed framework is extended to the registration of CT and US images, which is free from any data specific assumptions and has better generalization capability as compared to the existing rule-based/classical approaches.

cs.CV↗

arXiv2018

Deep Convolutional Neural Networks in the Face of Caricature: Identity and Image Revealed

Real-world face recognition requires an ability to perceive the unique features of an individual face across multiple, variable images. The primate visual system solves the problem of image invariance using cascades of neurons that convert images of faces into categorical representations of facial identity. Deep convolutional neural networks (DCNNs) also create generalizable face representations, but with cascades of simulated neurons. DCNN representations can be examined in a multidimensional "face space", with identities and image parameters quantified via their projections onto the axes that define the space. We examined the organization of viewpoint, illumination, gender, and identity in this space. We show that the network creates a highly organized, hierarchically nested, face similarity structure in which information about face identity and imaging characteristics coexist. Natural image variation is accommodated in this hierarchy, with face identity nested under gender, illumination nested under identity, and viewpoint nested under illumination. To examine identity, we caricatured faces and found that network identification accuracy increased with caricature level, and--mimicking human perception--a caricatured distortion of a face "resembled" its veridical counterpart. Caricatures improved performance by moving the identity away from other identities in the face space and minimizing the effects of illumination and viewpoint. Deep networks produce face representations that solve long-standing computational problems in generalized face recognition. They also provide a unitary theoretical framework for reconciling decades of behavioral and neural results that emphasized either the image or the object/face in representations, without understanding how a neural code could seamlessly accommodate both.

cs.CV↗

arXiv2018

A Proposal-Based Solution to Spatio-Temporal Action Detection in Untrimmed Videos

Existing approaches for spatio-temporal action detection in videos are limited by the spatial extent and temporal duration of the actions. In this paper, we present a modular system for spatio-temporal action detection in untrimmed security videos. We propose a two stage approach. The first stage generates dense spatio-temporal proposals using hierarchical clustering and temporal jittering techniques on frame-wise object detections. The second stage is a Temporal Refinement I3D (TRI-3D) network that performs action classification and temporal refinement on the generated proposals. The object detection-based proposal generation step helps in detecting actions occurring in a small spatial region of a video frame, while temporal jittering and refinement helps in detecting actions of variable lengths. Experimental results on the spatio-temporal action detection dataset - DIVA - show the effectiveness of our system. For comparison, the performance of our system is also evaluated on the THUMOS14 temporal action detection dataset.

cs.CV↗

arXiv2018

A Fast and Accurate System for Face Detection, Identification, and Verification

The availability of large annotated datasets and affordable computation power have led to impressive improvements in the performance of CNNs on various object detection and recognition benchmarks. These, along with a better understanding of deep learning methods, have also led to improved capabilities of machine understanding of faces. CNNs are able to detect faces, locate facial landmarks, estimate pose, and recognize faces in unconstrained images and videos. In this paper, we describe the details of a deep learning pipeline for unconstrained face identification and verification which achieves state-of-the-art performance on several benchmark datasets. We propose a novel face detector, Deep Pyramid Single Shot Face Detector (DPSSD), which is fast and capable of detecting faces with large scale variations (especially tiny faces). We give design details of the various modules involved in automatic face recognition: face detection, landmark localization and alignment, and face identification/verification. We provide evaluation results of the proposed face detector on challenging unconstrained face detection datasets. Then, we present experimental results for IARPA Janus Benchmarks A, B and C (IJB-A, IJB-B, IJB-C), and the Janus Challenge Set 5 (CS5).

cs.CV↗