arXiv subjects

Shubham Sharma

Publications and source records attributed to Shubham Sharma.

At least 73 records · Page 4Linked to original sources

arXiv2023

LINGO : Visually Debiasing Natural Language Instructions to Support Task Diversity

Cross-task generalization is a significant outcome that defines mastery in natural language understanding. Humans show a remarkable aptitude for this, and can solve many different types of tasks, given definitions in the form of textual instructions and a small set of examples. Recent work with pre-trained language models mimics this learning style: users can define and exemplify a task for the model to attempt as a series of natural language prompts or instructions. While prompting approaches have led to higher cross-task generalization compared to traditional supervised learning, analyzing 'bias' in the task instructions given to the model is a difficult problem, and has thus been relatively unexplored. For instance, are we truly modeling a task, or are we modeling a user's instructions? To help investigate this, we develop LINGO, a novel visual analytics interface that supports an effective, task-driven workflow to (1) help identify bias in natural language task instructions, (2) alter (or create) task instructions to reduce bias, and (3) evaluate pre-trained model performance on debiased task instructions. To robustly evaluate LINGO, we conduct a user study with both novice and expert instruction creators, over a dataset of 1,616 linguistic tasks and their natural language instructions, spanning 55 different languages. For both user groups, LINGO promotes the creation of more difficult tasks for pre-trained models, that contain higher linguistic diversity and lower instruction bias. We additionally discuss how the insights learned in developing and evaluating LINGO can aid in the design of future dashboards that aim to minimize the effort involved in prompt creation across multiple domains.

cs.HC↗

arXiv2023

Twist-4 T-even proton TMDs in the light-front quark-diquark model

We have dealt with the twist-4 T-even transverse momentum-dependent parton distributions (TMDs) for the case of proton in the light-front quark-diquark model (LFQDM). By decoding the unintegrated quark-quark correlator for the semi-inclusive deep inelastic scattering (SIDIS), we have specifically obtained the overlap form for the unpolarized \bigg($f_{3}^ν(x, {\bf p_\perp^2})$\bigg), longitudinally polarized \bigg($ g_{3L}^ν(x, {\bf p_\perp^2}),~h_{3L}^{\perpν}(x, {\bf p_\perp^2})$\bigg) and transversely polarized \bigg( ${g}^{ν}_{3T}(x, {\bf p_\perp^2}),~{h}^{ν}_{3T}(x, {\bf p_\perp^2})$ and ${h}^{ν\perp}_{3T}(x, {\bf p_\perp^2})$\bigg) proton TMDs. We have provided the explicit expressions for both the cases of the diquark being a scalar or a vector. Average transverse momenta and the average square transverse momenta for the TMDs have been calculated and the results have been tabulated with corresponding leading twist TMDs. In addition, the value of average transverse momentum and average square transverse momentum for TMD ${f}^{ν}_3(x, {\bf p_\perp^2})$ has been compared with the available light-front constituent quark model (LFCQM) results. From TMDs, we have also obtained and discussed the transverse momentum-dependent parton distribution functions (TMDPDFs). The model relations of the twist-4 T-even TMDs with the available leading twist T-even TMDs have also been obtained.

hep-ph↗

arXiv2022

Depth from Defocus Technique Applied to Unsteady Shock-Drop Secondary atomization

The two-sensor depth from defocus (DFD) technique for the measurement of drop sizes in a spray is further developed to achieve higher spatial and temporal resolution, to improve estimates of size and number concentration, and to provide additional guidelines for the calibration and design of the optical system for a specific application. The technique and these improvements are demonstrated using the case of secondary atomization when a shock wave interacts with a single drop. This is an application in which both high spatially and temporally resolved number density and size distributions of secondary droplets generated in the wake of the original drop are necessary.

physics.flu-dyn↗

arXiv2022

Hope Speech Detection on Social Media Platforms

Since personal computers became widely available in the consumer market, the amount of harmful content on the internet has significantly expanded. In simple terms, harmful content is anything online which causes a person distress or harm. It may include hate speech, violent content, threats, non-hope speech, etc. The online content must be positive, uplifting and supportive. Over the past few years, many studies have focused on solving this problem through hate speech detection, but very few focused on identifying hope speech. This paper discusses various machine learning approaches to identify a sentence as Hope Speech, Non-Hope Speech, or a Neutral sentence. The dataset used in the study contains English YouTube comments and is released as a part of the shared task "EACL-2021: Hope Speech Detection for Equality, Diversity, and Inclusion". Initially, the dataset obtained from the shared task had three classes: Hope Speech, non-Hope speech, and not in English; however, upon deeper inspection, we discovered that dataset relabeling is required. A group of undergraduates was hired to help perform the entire dataset's relabeling task. We experimented with conventional machine learning models (such as Naïve Bayes, logistic regression and support vector machine) and pre-trained models (such as BERT) on relabeled data. According to the experimental results, the relabeled data has achieved a better accuracy for Hope speech identification than the original data set.

cs.CL↗

arXiv2022

FASTER-CE: Fast, Sparse, Transparent, and Robust Counterfactual Explanations

Counterfactual explanations have substantially increased in popularity in the past few years as a useful human-centric way of understanding individual black-box model predictions. While several properties desired of high-quality counterfactuals have been identified in the literature, three crucial concerns: the speed of explanation generation, robustness/sensitivity and succinctness of explanations (sparsity) have been relatively unexplored. In this paper, we present FASTER-CE: a novel set of algorithms to generate fast, sparse, and robust counterfactual explanations. The key idea is to efficiently find promising search directions for counterfactuals in a latent space that is specified via an autoencoder. These directions are determined based on gradients with respect to each of the original input features as well as of the target, as estimated in the latent space. The ability to quickly examine combinations of the most promising gradient directions as well as to incorporate additional user-defined constraints allows us to generate multiple counterfactual explanations that are sparse, realistic, and robust to input manipulations. Through experiments on three datasets of varied complexities, we show that FASTER-CE is not only much faster than other state of the art methods for generating multiple explanations but also is significantly superior when considering a larger set of desirable (and often conflicting) properties. Specifically we present results across multiple performance metrics: sparsity, proximity, validity, speed of generation, and the robustness of explanations, to highlight the capabilities of the FASTER-CE family.

cs.LG↗

arXiv2022

FEAMOE: Fair, Explainable and Adaptive Mixture of Experts

Three key properties that are desired of trustworthy machine learning models deployed in high-stakes environments are fairness, explainability, and an ability to account for various kinds of "drift". While drifts in model accuracy, for example due to covariate shift, have been widely investigated, drifts in fairness metrics over time remain largely unexplored. In this paper, we propose FEAMOE, a novel "mixture-of-experts" inspired framework aimed at learning fairer, more explainable/interpretable models that can also rapidly adjust to drifts in both the accuracy and the fairness of a classifier. We illustrate our framework for three popular fairness measures and demonstrate how drift can be handled with respect to these fairness constraints. Experiments on multiple datasets show that our framework as applied to a mixture of linear experts is able to perform comparably to neural networks in terms of accuracy while producing fairer models. We then use the large-scale HMDA dataset and show that while various models trained on HMDA demonstrate drift with respect to both accuracy and fairness, FEAMOE can ably handle these drifts with respect to all the considered fairness measures and maintain model accuracy as well. We also prove that the proposed framework allows for producing fast Shapley value explanations, which makes computationally efficient feature attribution based explanations of model decisions readily available via FEAMOE.

cs.LG↗

arXiv2022

Regular charged black hole in massive gravity as heat engine

We examine the effect of addition of graviton mass on the efficiency of a regular charged black hole (RCB). In doing so, we make few comments on the critical behaviour of the system and calculate the relevant thermodynamic quantities such as entropy, Hawkings temperature and heat capacity. We confirm that graviton mass has a positive effect on the efficacy of such a heat engine under consideration.

gr-qc↗

arXiv2022

Quantum fluctuations of a regular charged black hole in massive gravity

We examine the thermodynamics of a regular charged black hole (RCB) added with corrections due to massive gravity and thermal fluctuations at quantum level. We then derive the expressions for all the relevant thermodynamic quantities such as entropy, Hawkings temperature, internal energy, Gibbs free energy, corrected to first order. We also briefly discuss the stability of such a system with the help of quantities like specific heat and thermodynamic pressure, along with this we present their comparative study with the corresponding original values.

gr-qc↗

arXiv2022

Shock induced aerobreakup of a polymeric droplet

Droplet atomization through aerobreakup is omnipresent in various natural and industrial processes. Atomization of Newtonian droplets is a well-studied area; however, non-Newtonian droplets have received less attention despite their frequent encounters. By subjecting polymeric droplets of different concentrations to the induced airflow behind a moving shock wave, we explore the role of elasticity in modulating the aerobreakup of viscoelastic droplets. Three distinct modes of aerobreakup are identified for a wide range of Weber number ($\sim 10^2-10^4$) and Elasticity number ($\sim 10^{-4}-10^2$) variation; these modes are: vibrational, shear-induced entrainment and catastrophic breakup mode. Each mode is described as a three stage process. Stage-I is the droplet deformation, stage-II is the appearance and growth of hydrodynamic instabilities, and stage-III is the evolution of liquid mass morphology. It is observed that elasticity plays an insignificant role in the first two stages, but a dominant role in the final stage. The results are described with the support of adequate mathematical analysis.

physics.flu-dyn↗

arXiv2022

Effect of Succinonitrile on Ion Transport in PEO-based Lithium Ion Battery Electrolytes

We report the ion transport mechanisms in succinonitrile (SN) loaded solid polymer electrolytes containing polyethylene oxide (PEO) and dissolved lithium bis(trifluoromethane)sulphonamide (LiTFSI) salt using molecular dynamics simulations. We investigated the effect of temperature and loading of SN on ion transport and relaxation phenomenon in PEO-LiTFSI electrolytes. It is observed that SN increases the ionic diffusivities in PEO-based solid polymer electrolytes and makes them suitable for battery applications. Interestingly, the diffusion coefficient of TFSI ions is an order of magnitude higher than the diffusion coefficient of lithium ions across the range of temperatures and loadings integrated. By analyzing different relaxation timescales and examining the underlying transport mechanisms in SN-loaded systems, we find that the diffusivity of TFSI ions correlates excellently with the Li-TFSI ion-pair relaxation timescales. In contrast, our simulations predict distinct transport mechanisms for Li-ions in SN-loaded PEO-LiTFSI electrolytes. Explicitly, the diffusivity of lithium ions cannot be uniquely determined by the ion-pair relaxation timescales but additionally depends on the polymer segmental dynamics. On the other hand, the SN loading induced diffusion coefficient at a given temperature does not correlate with either the ion-pair relaxation timescales or the polymer segmental relaxation timescales.

cond-mat.soft↗

arXiv2022

T-odd quark-gluon-quark correlation function in the light-front quark-diquark model

We have scrutinized the transverse momentum dependent quark-gluon-quark correlation function. We have utilized the light-front quark-diquark model to study the time-reversal-odd interaction-dependent twist-3 gluon distributions which we have obtained from the disintegration of the transverse momentum dependent quark-gluon-quark correlator. Specifically, we have studied the behavior of $\tilde{e}_{L}$ and $\tilde{e}_{T}$ while considering our diquark to be an axial-vector.

hep-ph↗

arXiv2020

FaiR-N: Fair and Robust Neural Networks for Structured Data

Fairness in machine learning is crucial when individuals are subject to automated decisions made by models in high-stake domains. Organizations that employ these models may also need to satisfy regulations that promote responsible and ethical A.I. While fairness metrics relying on comparing model error rates across subpopulations have been widely investigated for the detection and mitigation of bias, fairness in terms of the equalized ability to achieve recourse for different protected attribute groups has been relatively unexplored. We present a novel formulation for training neural networks that considers the distance of data points to the decision boundary such that the new objective: (1) reduces the average distance to the decision boundary between two groups for individuals subject to a negative outcome in each group, i.e. the network is more fair with respect to the ability to obtain recourse, and (2) increases the average distance of data points to the boundary to promote adversarial robustness. We demonstrate that training with this loss yields more fair and robust neural networks with similar accuracies to models trained without it. Moreover, we qualitatively motivate and empirically show that reducing recourse disparity across groups also improves fairness measures that rely on error rates. To the best of our knowledge, this is the first time that recourse capabilities across groups are considered to train fairer neural networks, and a relation between error rates based fairness and recourse based fairness is investigated.

cs.LG↗

arXiv2020

On Secondary Atomization and blockage of surrogate cough droplets in single and multi-layer face masks

By now it is well-understood that the usage of facemasks provides protection from transmission of viral loads through exhalation and inhalation of respiratory droplets. Therefore, during the current Covid-19 pandemic the usage of face masks is strongly recommended by health officials. Although three-layer masks are generally advised for usage, many commonly available or homemade masks contain only single and double layers. In this study, we show through detailed physics based analyses and high speed imaging that high momentum cough droplets on impingement on single- and double-layer masks can lead to significant partial penetration and more importantly atomization into numerous much smaller daughter droplets, thereby increasing the total population of the aerosol, which can remain suspended for a longer time. The possibility of secondary atomization of high momentum cough droplets due to impingement, hydrodynamic focusing and extrusion through the microscale pores in the fibrous network of the mask has not been explored before. However, this unique mode of aerosol generation poses a finite risk of infection as shown in this work. We also demonstrate that in single layer masks close to 70 % of a given droplet volume is atomized and only 30 % is trapped within the fibers. The entrapped volume is close to 90 % for double layer masks which still allows some atomization into smaller droplets. We however found that a triple-layer surgical mask permits negligible penetration and hence should be effective in preventing disease transmission.

physics.app-ph↗

arXiv2020

Explainable Machine Learning in Deployment

Explainable machine learning offers the potential to provide stakeholders with insights into model behavior by using various methods such as feature importance scores, counterfactual explanations, or influential training data. Yet there is little understanding of how organizations use these methods in practice. This study explores how organizations view and use explainability for stakeholder consumption. We find that, currently, the majority of deployments are not for end users affected by the model but rather for machine learning engineers, who use explainability to debug the model itself. There is thus a gap between explainability in practice and the goal of transparency, since explanations primarily serve internal stakeholders rather than external ones. Our study synthesizes the limitations of current explainability techniques that hamper their use for end users. To facilitate end user interaction, we develop a framework for establishing clear goals for explainability. We end by discussing concerns raised regarding explainability.

cs.LG↗

arXiv2020

Reliable computational prediction of supramolecular ordering of complex molecules under electrochemical conditions

We propose a computationally lean, two-stage approach that reliably predicts self-assembly behavior of complex charged molecules on a metallic surfaces under electrochemical conditions. Stage one uses ab initio simulations to provide reference data for the energies (evaluated for archetypical configurations) to fit the parameters of a conceptually much simpler and computationally less expensive model of the molecules: classical, spherical particles, representing the respective atomic entities, a soft but perfectly conductive wall potential represents the metallic surface. Stage two feeds the energies that emerge from this classical model into highly efficient and reliable optimization techniques to identify via energy minimization the ordered ground state configurations of the molecules. We demonstrate the power of our approach by successfully reproducing, on a semi-quantitative level, the intricate supramolecular ordering observed experimentally for PQP$^+$ and ClO$_4^-$ molecules at an Au(111)-electrolyte interface, including the formation of open-porous, self-hosts--guest, and stratified bilayer phases as a function of the electric field at the solid--liquid interface. We also discuss the role of the perchlorate ions in the self-assembly process, whose positions could not be identified in the related experimental investigations.

physics.comp-ph↗

arXiv2020

Development of On Board Computer for a Nanosatellite

This paper describes the process of conceptualization, design, and testing of the Onboard Computer (OBC) Software for a 3U nanosatellite. The on-board computer of the satellite is responsible for initiating dataflow between onboard hardware, performing image compression, and run control algorithms like fine pointing, sun pointing, ground pointing for payload operation and idle state detumbling. The actuation is carried out by interfacing magnetorquers and reaction wheels with the OBC. The software of the onboard computer is implemented on a Linux based operating system run on the ARM Cortex A9 processor which is part of the Zynq-7000 SoC. A Field Programmable Gate Array (FPGA) is used specifically for image compression. The compressed image is stored in a serial flash memory shared between the camera and the FPGA. The architecture comprises of a system-wide I2C bus to which the sensors interfaced. The collected data is used for logging followed by downlink and as input to algorithms used for pointing and detumbling. An SPI interface is used between the Power Subsystem microcontroller and the On-Board Computer since a large amount of housekeeping data will have to be exchanged at high rates. Reaction wheels and magnetorquers are actuated by current driver circuits which get the control signals from the OBC. The satellite is modelled as a Finite State Machine for software development. The states broadly fall under two categories, Normal and Emergency. Each state has a predetermined set of logical tasks to be run, which are abstracted as separate processes in the memory. State transitions take place by polling the health metrics of the satellite. However, hardware interrupts are implemented on selected peripherals which ensure an asynchronous switching to the Emergency States for safety. A review of some common fault detection, isolation and removal methods used shall conclude the paper.

eess.SY↗

arXiv2019

CERTIFAI: Counterfactual Explanations for Robustness, Transparency, Interpretability, and Fairness of Artificial Intelligence models

As artificial intelligence plays an increasingly important role in our society, there are ethical and moral obligations for both businesses and researchers to ensure that their machine learning models are designed, deployed, and maintained responsibly. These models need to be rigorously audited for fairness, robustness, transparency, and interpretability. A variety of methods have been developed that focus on these issues in isolation, however, managing these methods in conjunction with model development can be cumbersome and timeconsuming. In this paper, we introduce a unified and model-agnostic approach to address these issues: Counterfactual Explanations for Robustness, Transparency, Interpretability, and Fairness of Artificial Intelligence models (CERTIFAI). Unlike previous methods in this domain, CERTIFAI is a general tool that can be applied to any black-box model and any type of input data. Given a model and an input instance, CERTIFAI uses a custom genetic algorithm to generate counterfactuals: instances close to the input that change the prediction of the model. We demonstrate how these counterfactuals can be used to examine issues of robustness, interpretability, transparency, and fairness. Additionally, we introduce CERScore, the first black-box model robustness score that performs comparably to methods that have access to model internals.

cs.LG↗

arXiv2019

Verity: Blockchains to Detect Insider Attacks in DBMS

Integrity and security of the data in database systems are typically maintained with access control policies and firewalls. However, insider attacks -- where someone with an intimate knowledge of the system and administrative privileges tampers with the data -- pose a unique challenge. Measures like append only logging prove to be insufficient because an attacker with administrative privileges can alter logs and login records to eliminate the trace of attack, thus making insider attacks hard to detect. In this paper, we propose Verity -- first of a kind system to the best of our knowledge. Verity serves as a dataless framework by which any blockchain network can be used to store fixed-length metadata about tuples from any SQL database, without complete migration of the database. Verity uses a formalism for parsing SQL queries and query results to check the respective tuples' integrity using blockchains to detect insider attacks. We have implemented our technique using Hyperledger Fabric, Composer REST API, and SQLite database. Using TPC-H data and SQL queries of varying complexity and types, our experiments demonstrate that any overhead of integrity checking remains constant per tuple in a query's results, and scales linearly.

cs.DB↗