arXiv subjects

Aditya Kulkarni

Publications and source records attributed to Aditya Kulkarni.

At least 19 recordsLinked to original sources

arXiv2026

The onset of the strong backreaction regime in axion inflation, an analytical study

Models where an axion-like inflaton is coupled to a $U(1)$ gauge field are theoretically well motivated and can display a rich phenomenology. The regime in which the quanta of the gauge field strongly backreact on the rolling inflaton has been studied for well more than a decade, and yet we cannot say that it is fully understood. In this paper we present an analytical study of the onset of the strong backreaction regime. Our formalism relies on a Laplace transform to convert a complicated integro-differential equation into the search of the poles of an analytic function. We confirm the existence, previously observed using a completely different formalism, of a stable region of strong backreaction. Our work provides, for the first time, analytical formulae that allow to study the evolution of the system as it transitions from the weak to the strong backreaction regime.

astro-ph.CO↗

arXiv2025

ODoQ: Oblivious DNS-over-QUIC

The Domain Name System (DNS), which converts domain names to their respective IP addresses, has advanced enhancements aimed at safeguarding DNS data and users' identity from attackers. The recent privacy-focused advancements have enabled the IETF to standardize several protocols. Nevertheless, these protocols tend to focus on either strengthening user privacy (like Oblivious DNS and Oblivious DNS-over-HTTPS) or reducing resolution latency (as demonstrated by DNS-over-QUIC). Achieving both within a single protocol remains a key challenge, which we address in this paper. Our proposed protocol -- 'Oblivious DNS-over-QUIC' (ODoQ) -- leverages the benefits of the QUIC protocol and incorporates an intermediary proxy server to protect the client's identity from exposure to the recursive resolver.

cs.CR↗

arXiv2025

Bridging the Gap in Phishing Detection: A Comprehensive Phishing Dataset Collector

To combat phishing attacks -- aimed at luring web users to divulge their sensitive information -- various phishing detection approaches have been proposed. As attackers focus on devising new tactics to bypass existing detection solutions, researchers have adapted by integrating machine learning and deep learning into phishing detection. Phishing dataset collection is vital to developing effective phishing detection approaches, which highly depend on the diversity of the gathered datasets. The lack of diversity in the dataset results in a biased model. Since phishing websites are often short-lived, collecting them is also a challenge. Consequently, very few phishing webpage dataset repositories exist to date. No single repository comprehensively consolidates all phishing elements corresponding to a phishing webpage, namely, URL, webpage source code, screenshot, and related webpage resources. This paper introduces a resource collection tool designed to gather various resources associated with a URL, such as CSS, Javascript, favicons, webpage images, and screenshots. Our tool leverages PhishTank as the primary source for obtaining active phishing URLs. Our tool fetches several additional webpage resources compared to PyWebCopy Python library, which provides webpage content for a given URL. Additionally, we share a sample dataset generated using our tool comprising 4,056 legitimate and 5,666 phishing URLs along with their associated resources. We also remark on the top correlated phishing features with their associated class label found in our dataset. Our tool offers a comprehensive resource set that can aid researchers in developing effective phishing detection approaches.

cs.CR↗

arXiv2025

Overcoming DNSSEC Islands of Security: A TLS and IP-Based Certificate Solution

The Domain Name System (DNS) serves as the backbone of the Internet, primarily translating domain names to IP addresses. Over time, various enhancements have been introduced to strengthen the integrity of DNS. Among these, DNSSEC stands out as a leading cryptographic solution. It protects against attacks (such as DNS spoofing) by establishing a chain of trust throughout the DNS nameserver hierarchy. However, DNSSEC's effectiveness is compromised when there is a break in this chain, resulting in "Islands of Security", where domains can authenticate locally but not across hierarchical levels, leading to a loss of trust and validation between them. Leading approaches to addressing these issues were centralized, with a single authority maintaining some kind of bulletin board. This approach requires significantly more infrastructure and places excessive trust in the entity responsible for managing it properly. In this paper, we propose a decentralized approach to addressing gaps in DNSSEC's chain of trust, commonly referred to as "Islands of Security". We leverage TLS and IP-based certificates to enable end-to-end authentication between hierarchical levels, eliminating the need for uniform DNSSEC deployment across every level of the DNS hierarchy. This approach enhances the overall integrity of DNSSEC, while reducing dependence on registrars for maintaining signature records to verify the child nameserver's authenticity. By offering a more flexible and efficient solution, our method strengthens DNS security and streamlines deployment across diverse environments.

cs.CR↗

arXiv2025

Phish-Blitz: Advancing Phishing Detection with Comprehensive Webpage Resource Collection and Visual Integrity Preservation

Phishing attacks are increasingly prevalent, with adversaries creating deceptive webpages to steal sensitive information. Despite advancements in machine learning and deep learning for phishing detection, attackers constantly develop new tactics to bypass detection models. As a result, phishing webpages continue to reach users, particularly those unable to recognize phishing indicators. To improve detection accuracy, models must be trained on large datasets containing both phishing and legitimate webpages, including URLs, webpage content, screenshots, and logos. However, existing tools struggle to collect the required resources, especially given the short lifespan of phishing webpages, limiting dataset comprehensiveness. In response, we introduce Phish-Blitz, a tool that downloads phishing and legitimate webpages along with their associated resources, such as screenshots. Unlike existing tools, Phish-Blitz captures live webpage screenshots and updates resource file paths to maintain the original visual integrity of the webpage. We provide a dataset containing 8,809 legitimate and 5,000 phishing webpages, including all associated resources. Our dataset and tool are publicly available on GitHub, contributing to the research community by offering a more complete dataset for phishing detection.

cs.CR↗

arXiv2025

Phishing Webpage Detection: Unveiling the Threat Landscape and Investigating Detection Techniques

In the realm of cybersecurity, phishing stands as a prevalent cyber attack, where attackers employ various tactics to deceive users into gathering their sensitive information, potentially leading to identity theft or financial gain. Researchers have been actively working on advancing phishing webpage detection approaches to detect new phishing URLs, bolstering user protection. Nonetheless, the ever-evolving strategies employed by attackers, aimed at circumventing existing detection approaches and tools, present an ongoing challenge to the research community. This survey presents a systematic categorization of diverse phishing webpage detection approaches, encompassing URL-based, webpage content-based, and visual techniques. Through a comprehensive review of these approaches and an in-depth analysis of existing literature, our study underscores current research gaps in phishing webpage detection. Furthermore, we suggest potential solutions to address some of these gaps, contributing valuable insights to the ongoing efforts to combat phishing attacks.

cs.CR↗

arXiv2025

IndicSQuAD: A Comprehensive Multilingual Question Answering Dataset for Indic Languages

The rapid progress in question-answering (QA) systems has predominantly benefited high-resource languages, leaving Indic languages largely underrepresented despite their vast native speaker base. In this paper, we present IndicSQuAD, a comprehensive multi-lingual extractive QA dataset covering nine major Indic languages, systematically derived from the SQuAD dataset. Building on previous work with MahaSQuAD for Marathi, our approach adapts and extends translation techniques to maintain high linguistic fidelity and accurate answer-span alignment across diverse languages. IndicSQuAD comprises extensive training, validation, and test sets for each language, providing a robust foundation for model development. We evaluate baseline performances using language-specific monolingual BERT models and the multilingual MuRIL-BERT. The results indicate some challenges inherent in low-resource settings. Moreover, our experiments suggest potential directions for future work, including expanding to additional languages, developing domain-specific datasets, and incorporating multimodal data. The dataset and models are publicly shared at https://github.com/l3cube-pune/indic-nlp

cs.CL↗

arXiv2025

Differentially Private Geodesic Regression

In statistical applications it has become increasingly common to encounter data structures that live on non-linear spaces such as manifolds. Classical linear regression, one of the most fundamental methodologies of statistical learning, captures the relationship between an independent variable and a response variable which both are assumed to live in Euclidean space. Thus, geodesic regression emerged as an extension where the response variable lives on a Riemannian manifold. The parameters of geodesic regression, as with linear regression, capture the relationship of sensitive data and hence one should consider the privacy protection practices of said parameters. We consider releasing Differentially Private (DP) parameters of geodesic regression via the K-Norm Gradient (KNG) mechanism for Riemannian manifolds. We derive theoretical bounds for the sensitivity of the parameters showing they are tied to their respective Jacobi fields and hence the curvature of the space. This corroborates, and extends, recent findings of differential privacy for the Fr\'echet mean. We demonstrate the efficacy of our methodology on the sphere, $S_2\subset\mathbb{R}^3$, the space of symmetric positive definite matrices, and Kendall's planar shape space. Our methodology is general to any Riemannian manifold, and thus it is suitable for data in domains such as medical imaging and computer vision.

stat.ML↗

arXiv2024

Spectra of fermions produced by a time-dependent axion in the radiation- and matter-dominated Universe

Axion-like degrees of freedom generally interact with fermions through a shift symmetric coupling. As a consequence, a time-dependent axion will lead to the generation of fermions by amplifying their vacuum fluctuations. We provide the formulae that allow one to determine the spectra of produced fermions in a generic Friedmann-Lemaitre-Robertson-Walker Universe with flat spatial slices. Then we derive simple approximate formulae for the spectra of the produced fermions, as a function of the model parameters, in the specific cases of a radiation- and a matter-dominated Universe, in the regime in which the backreaction of the produced fermions on the axionic background can be neglected.

astro-ph.CO↗

arXiv2024

From ML to LLM: Evaluating the Robustness of Phishing Webpage Detection Models against Adversarial Attacks

Phishing attacks attempt to deceive users into stealing sensitive information, posing a significant cybersecurity threat. Advances in machine learning (ML) and deep learning (DL) have led to the development of numerous phishing webpage detection solutions, but these models remain vulnerable to adversarial attacks. Evaluating their robustness against adversarial phishing webpages is essential. Existing tools contain datasets of pre-designed phishing webpages for a limited number of brands, and lack diversity in phishing features. To address these challenges, we develop PhishOracle, a tool that generates adversarial phishing webpages by embedding diverse phishing features into legitimate webpages. We evaluate the robustness of three existing task-specific models - Stack model, VisualPhishNet, and Phishpedia - against PhishOracle-generated adversarial phishing webpages and observe a significant drop in their detection rates. In contrast, a multimodal large language model (MLLM)-based phishing detector demonstrates stronger robustness against these adversarial attacks but still is prone to evasion. Our findings highlight the vulnerability of phishing detection models to adversarial attacks, emphasizing the need for more robust detection approaches. Furthermore, we conduct a user study to evaluate whether PhishOracle-generated adversarial phishing webpages can deceive users. The results show that many of these phishing webpages evade not only existing detection models but also users.

cs.CR↗

arXiv2024

MahaSQuAD: Bridging Linguistic Divides in Marathi Question-Answering

Question-answering systems have revolutionized information retrieval, but linguistic and cultural boundaries limit their widespread accessibility. This research endeavors to bridge the gap of the absence of efficient QnA datasets in low-resource languages by translating the English Question Answering Dataset (SQuAD) using a robust data curation approach. We introduce MahaSQuAD, the first-ever full SQuAD dataset for the Indic language Marathi, consisting of 118,516 training, 11,873 validation, and 11,803 test samples. We also present a gold test set of manually verified 500 examples. Challenges in maintaining context and handling linguistic nuances are addressed, ensuring accurate translations. Moreover, as a QnA dataset cannot be simply converted into any low-resource language using translation, we need a robust method to map the answer translation to its span in the translated passage. Hence, to address this challenge, we also present a generic approach for translating SQuAD into any low-resource language. Thus, we offer a scalable approach to bridge linguistic and cultural gaps present in low-resource languages, in the realm of question-answering systems. The datasets and models are shared publicly at https://github.com/l3cube-pune/MarathiNLP .

cs.CL↗

arXiv2024

Mitigating Bias in Machine Learning Models for Phishing Webpage Detection

The widespread accessibility of the Internet has led to a surge in online fraudulent activities, underscoring the necessity of shielding users' sensitive information from cybercriminals. Phishing, a well-known cyberattack, revolves around the creation of phishing webpages and the dissemination of corresponding URLs, aiming to deceive users into sharing their sensitive information, often for identity theft or financial gain. Various techniques are available for preemptively categorizing zero-day phishing URLs by distilling unique attributes and constructing predictive models. However, these existing techniques encounter unresolved issues. This proposal delves into persistent challenges within phishing detection solutions, particularly concentrated on the preliminary phase of assembling comprehensive datasets, and proposes a potential solution in the form of a tool engineered to alleviate bias in ML models. Such a tool can generate phishing webpages for any given set of legitimate URLs, infusing randomly selected content and visual-based phishing features. Furthermore, we contend that the tool holds the potential to assess the efficacy of existing phishing detection solutions, especially those trained on confined datasets.

cs.CR↗

arXiv2023

STEER: Semantic Turn Extension-Expansion Recognition for Voice Assistants

In the context of a voice assistant system, steering refers to the phenomenon in which a user issues a follow-up command attempting to direct or clarify a previous turn. We propose STEER, a steering detection model that predicts whether a follow-up turn is a user's attempt to steer the previous command. Constructing a training dataset for steering use cases poses challenges due to the cold-start problem. To overcome this, we developed heuristic rules to sample opt-in usage data, approximating positive and negative samples without any annotation. Our experimental results show promising performance in identifying steering intent, with over 95% accuracy on our sampled data. Moreover, STEER, in conjunction with our sampling strategy, aligns effectively with real-world steering scenarios, as evidenced by its strong zero-shot performance on a human-graded evaluation set. In addition to relying solely on user transcripts as input, we introduce STEER+, an enhanced version of the model. STEER+ utilizes a semantic parse tree to provide more context on out-of-vocabulary words, such as named entities that often occur at the sentence boundary. This further improves model performance, reducing error rate in domains where entities frequently appear, such as messaging. Lastly, we present a data analysis that highlights the improvement in user experience when voice assistants support steering use cases.

cs.CL↗

arXiv2023

Advancing Diagnostic Precision: Leveraging Machine Learning Techniques for Accurate Detection of Covid-19, Pneumonia, and Tuberculosis in Chest X-Ray Images

Lung diseases such as COVID-19, tuberculosis (TB), and pneumonia continue to be serious global health concerns that affect millions of people worldwide. In medical practice, chest X-ray examinations have emerged as the norm for diagnosing diseases, particularly chest infections such as COVID-19. Paramedics and scientists are working intensively to create a reliable and precise approach for early-stage COVID-19 diagnosis in order to save lives. But with a variety of symptoms, medical diagnosis of these disorders poses special difficulties. It is essential to address their identification and timely diagnosis in order to successfully treat and prevent these illnesses. In this research, a multiclass classification approach using state-of-the-art methods for deep learning and image processing is proposed. This method takes into account the robustness and efficiency of the system in order to increase diagnostic precision of chest diseases. A comparison between a brand-new convolution neural network (CNN) and several transfer learning pre-trained models including VGG19, ResNet, DenseNet, EfficientNet, and InceptionNet is recommended. Publicly available and widely used research datasets like Shenzen, Montogomery, the multiclass Kaggle dataset and the NIH dataset were used to rigorously test the model. Recall, precision, F1-score, and Area Under Curve (AUC) score are used to evaluate and compare the performance of the proposed model. An AUC value of 0.95 for COVID-19, 0.99 for TB, and 0.98 for pneumonia is obtained using the proposed network. Recall and precision ratings of 0.95, 0.98, and 0.97, respectively, likewise met high standards.

eess.IV↗

arXiv2023

Businesses in high-income zip codes often saw sharper visit reductions during the COVID-19 pandemic

As the COVID-19 pandemic unfolded, the mobility patterns of people worldwide changed drastically. While travel time, costs, and trip convenience have always influenced mobility, the risk of infection and policy actions such as lockdowns and stay-at-home orders emerged as new factors to consider in the location-visitation calculus. We use SafeGraph mobility data from Minnesota, USA, to demonstrate that businesses (especially those requiring extended indoor visits) located in affluent zip codes witnessed sharper reductions in visits (relative to pre-pandemic times) outside of the lockdown periods than their poorer counterparts. To the extent visits translate into sales, we contend that post-pandemic recovery efforts should prioritize relief funding, keeping the losses relating to diminished visits in mind.

cs.CY↗

arXiv2022

Understanding COVID-19 Effects on Mobility: A Community-Engaged Approach

Given aggregated mobile device data, the goal is to understand the impact of COVID-19 policy interventions on mobility. This problem is vital due to important societal use cases, such as safely reopening the economy. Challenges include understanding and interpreting questions of interest to policymakers, cross-jurisdictional variability in choice and time of interventions, the large data volume, and unknown sampling bias. The related work has explored the COVID-19 impact on travel distance, time spent at home, and the number of visitors at different points of interest. However, many policymakers are interested in long-duration visits to high-risk business categories and understanding the spatial selection bias to interpret summary reports. We provide an Entity Relationship diagram, system architecture, and implementation to support queries on long-duration visits in addition to fine resolution device count maps to understand spatial bias. We closely collaborated with policymakers to derive the system requirements and evaluate the system components, the summary reports, and visualizations.

cs.CY↗

arXiv2021

VLSI Systems for signal processing and Communications

The growing advances in VLSI technology and design tools have exponentially expanded the application domain of digital signal processing over the past 10 years. This survey emphasises on the architectural and performance parameters of VLSI for DSP applications such as speech processing, wireless communication, analog to digital converters, etc

eess.SP↗

arXiv2020

Semantic Segmentation of Medium-Resolution Satellite Imagery using Conditional Generative Adversarial Networks

Semantic segmentation of satellite imagery is a common approach to identify patterns and detect changes around the planet. Most of the state-of-the-art semantic segmentation models are trained in a fully supervised way using Convolutional Neural Network (CNN). The generalization property of CNN is poor for satellite imagery because the data can be very diverse in terms of landscape types, image resolutions, and scarcity of labels for different geographies and seasons. Hence, the performance of CNN doesn't translate well to images from unseen regions or seasons. Inspired by Conditional Generative Adversarial Networks (CGAN) based approach of image-to-image translation for high-resolution satellite imagery, we propose a CGAN framework for land cover classification using medium-resolution Sentinel-2 imagery. We find that the CGAN model outperforms the CNN model of similar complexity by a significant margin on an unseen imbalanced test dataset.

cs.CV↗