Browsing by Author "Tiwari, Kamlesh"

Now showing 1 - 20 of 42

Bayesian deep learning meets self-attention: a risk-aware approach to advertisement optimization
(IEEE, 2025-05) Bhatia, Ashutosh; Tiwari, Kamlesh
In the highly competitive landscape of e-commerce advertising, maximizing Return on Advertising Spend (ROAS) is critical, yet remains inherently uncertain due to auction-based bidding dynamics and fluctuating market conditions. Traditional deterministic models fail to capture this uncertainty, necessitating a probabilistic approach that balances predictive accuracy with interpretability. To address this challenge, the paper proposes a novel Hierarchical Bayesian Deep Learning framework that integrates a Bayesian Belief Network (BBN) for structured probabilistic reasoning and a Mixture Density Network (MDN) for full distributional modeling of ROAS. The BBN models dependencies among campaign variables, offering interpretable insights, while the hierarchical deep learning architecture overcomes scalability limitations in high-dimensional settings through self-attention mechanisms. Experiments demonstrate up to 22.8% lower RMSE and 27.4% better Negative Log Likelihood (NLL) and up to 31.2% lower Kullback-Leibler divergence (KLD) than state-of-the-art methods (DeepAR, Prophet, NGBoost), achieving an R2 of 98% with an inference speed of 5.2 ms per campaign, making real-time bidding feasible. Ablation studies confirm that attention-driven feature selection and calibrated uncertainty quantification significantly enhance both predictive performance and explainability, identifying key drivers of campaign success. By providing precise, uncertainty-aware, and explainable predictions, this approach enables adaptive bidding strategies, optimized budget allocation, and risk management, setting a new benchmark for intelligent decision-making in digital advertising.
Bitcoin’s Blockchain Data Analytics: A Graph Theoretic Perspective
(Springer, 2022-03) Bhatia, Ashutosh; Tiwari, Kamlesh
Bitcoin is the first and most widely used cryptocurrency in the world. It provides a pseudonym identity to its users that is established using the user’s public key, which leads to preserving the user’s privacy. Each transfer of bitcoin cryptocurrency among the users makes a transaction. The pseudonym identities are considered as transaction end-points. These transactions are recorded on an immutable public ledger called Blockchain which is an append-only data structure. The popularity of Bitcoin has increased unreasonably. The general trend shows a positive response from the common masses indicating an increase in trust and privacy concerns which makes an interesting use case from the analysis point of view. Moreover, since the blockchain is publicly available and up-to-date, any analysis would provide a live insight into the usage patterns which ultimately would be useful for making a number of inferences by law-enforcement agencies, economists, tech-enthusiasts, etc. In this paper, we study various applications and techniques of performing data analytics over Bitcoin blockchain from a graph theoretic perspective. We also propose a framework for performing such data analytics and explored a couple of use cases using the proposed framework.
CP-Net: Multi-Scale Core Point Localization in Fingerprints Using Hourglass Network
(IEEE, 2023) Bhatia, Ashutosh; Tiwari, Kamlesh
Core point is a location that exhibits high curvature properties in a fingerprint. Detecting the accurate location of a core point is useful for efficient fingerprint matching, classification, and identification tasks. This paper proposes CP-Net, a novel core point detection network that comprises the Macro Localization Network (MLN) and the Micro-Regression Network (MRN). MLN is a specialized autoencoder network with an hourglass network at its bottleneck. It takes an input fingerprint image and outputs a region of interest that could be the most probable region containing the core point. The second component, MRN, regresses the RoI and locates the coordinates of the core point in the given fingerprint sample. Introducing an hourglass network in the MLN bottleneck ensures multi-scale spatial attention that captures local and global contexts and facilitates a higher localization accuracy for that area. Unlike existing multi-stage models, the components are stacked and trained in an end-to-end manner. Experiments have been performed on three widely used publicly available fingerprint datasets, namely, FVC2002 DB1A, FVC2004 DB1A, and FVC2006 DB2A. The proposed model achieved a true detection rate (TDR) of 98%, 100%, and 99.04% respectively, while considering 20 pixels distance from the ground truth as correct. Obtained experimental results on the considered datasets demonstrate that CP-Net outperforms the state-of-the-art core point detection techniques.
D-insta: A Decentralized Image Sharing Platform
(Springer, 2023-03) Bhatia, Ashutosh; Tiwari, Kamlesh
Due to the covid-19 pandemic, people have moved toward digitization and using digital technologies in their daily life. For instance, photographers and artists use social media platforms or stock photo websites to showcase their art to people to get recognition and credit. Since social media platforms attract people more than stock photo websites, we consider incorporating the stock photo website features into the social media platforms. Currently, such platforms are running in a centralized fashion where their proprietary algorithms mask most of the content to which some users and advertisement posts are given more priority. Due to the centralization, such hidden algorithms create trust issues among the users along with other issues such as single point of failure, identity theft, etc. This causes genuine artists and photographers to lose their interest and motivation. Providing due credit to the authors and deserved recognition are significant concerns for photographers who share images on stock photo websites or social media platforms. In this paper, we propose a decentralized image-sharing platform/application utilizing blockchain and a distributed file storage system to address all these issues. The proposed platform leverages Ethereum-based smart contracts to maintain trust as deployed smart contracts are immutable, and the logic written in them is publicly available. We leverage a distributed file storage system to solve the blockchain scalability issue in terms of storage.
DCGit: Decentralized Internet Hosting for Software Development
(IEEE, 2023) Bhatia, Ashutosh; Tiwari, Kamlesh
Git has been the de-facto version control system for the Software Development industry. Although Git is distributed, developers’ tools for collaboration, such as GitHub, are centralized entities owned by large corporations such as Microsoft. The centralization creates trust and privacy issues for software development companies (preserving their intellectual property), along with a significant" single point of failure" issue. In addition, such centralized systems are susceptible to Sybil and distributed denial of service (DDoS) attacks due to the presence of malicious individuals. Blockchain technology has many key characteristics (such as decentralization, transparency, immutability, and audibility), solving these centralization issues. However, the requirement of having a storage system to store the user’s repositories over the blockchain creates a scalability issue (in terms of storage). Most importantly, it makes data (code) privacy more severe due to its open nature. In this paper, we propose a privacy-preserving decentralized alternative solution and framework named "DCGit" powered by Web3 technologies such as the Ethereum Blockchain and InterPlanetary File System (IPFS) to provide security and scalability yet user-friendly collaboration for software development.
Decentralized marketplace for maintenance of electric vehicles
(Springer, 2025-04) Bhatia, Ashutosh; Tiwari, Kamlesh
As electric vehicles (EVs) become an integral part of the global transportation ecosystem, the need for efficient and cost-effective maintenance solutions will rise. This paper explores the application of game theory, specifically reverse auctions, to establish a decentralized, driver-centric marketplace for EV maintenance, promoting sustainability. The proposed system enables EV drivers to act as price determiners by launching individual smart contracts, which serve as reverse auction platforms. Maintenance providers can then bid on these contracts, competing to offer the most cost-effective services. By leveraging blockchain technology, smart contracts ensure transparency, trust, and secure fund management throughout the process. This innovative approach empowers EV drivers, reduces maintenance costs, and promotes a competitive service market. The paper outlines the underlying mechanisms, system architecture, and potential benefits of this model, along with a discussion of its implementation challenges and future implications for the EV ecosystem.
Deep learning approaches for driver distraction detection using driver facing cameras: literature review and empirical study using cnn classifiers on a 100-driver image dataset
(2025-05) Bhatia, Ashutosh; Sharma, Yashvardhan; Tiwari, Kamlesh
Distracted driving contributes to thousands of fatalities and injuries globally. According to India’s Ministry of Road Transport and Highways (MoRTH), distraction-related behaviors such as rear-end and off-road collisions accounted for nearly one-fourth of all traffic incidents in 2022. The U.S. National Highway Traffic Safety Administration (NHTSA) reported 3,275 deaths and over 324,000 injuries from distraction-related crashes in 2023. In Europe, the European Road Safety Observatory (ERSO) observed handheld phone use by drivers in up to 9.4% of vehicles across member states, with self-reported texting rates reaching 53%. Despite numerous studies and surveys on driver distraction detection, existing literature remains fragmented, often combining multiple sensor modalities or distraction with related driver states such as fatigue. Prior empirical efforts also lack a unified benchmarking strategy to assess model generalization under shifts in viewpoint or spectral input. This paper presents a focused survey and empirical study of visiononly distraction detection using deep learning models applied to driver-facing camera inputs. It introduces a conceptual model linking behavioral cues to cognitive distraction, defines the visionbased Driver Distraction Detection (vDDD) system with alert logic, and develops structured taxonomies of datasets, architectures, and learning strategies. Using the 100-Driver dataset, the empirical study evaluates 26 CNN classifiers under 64 crossdomain configurations, systematically analyzing generalization across modality and camera view changes. Results show that frontal RGB-trained models generalize better than their NIRtrained counterparts and that lightweight models trade off accuracy under rare class scenarios for faster inference. The study establishes the vDDD paradigm as a vision-based behavioral modeling approach for distraction detection using driver-facing camera data. It outlines future research directions in spectrumaligned augmentation, attention modeling, and lightweight visuallanguage fusion, emphasizing deployment-focused strategies such as quantization, contrastive learning, and progressive fine-tuning.
DTeeth: Teeth-photo Based Human Authentication for Mobile Devices
(IEEE, 2022) Bhatia, Ashutosh; Tiwari, Kamlesh
This paper investigates teeth-photo, a new biometric modality, for human authentication on mobile and hand-held devices. The proposed system is suitable for multiple applications including device unlocking and secure authentication. Teeth samples have been acquired using a mobile application having markers to register the teeth area. The region of interest (RoI) is then extracted using the markers and the same is enhanced for better visual clarity. A deep learning architecture along with the feature regularization scheme is devised to obtain highly discriminative embedding. The model is trained in an end-to-end manner with a few samples and thus, is efficient in terms of time and energy requirements. Experiments have been conducted on an in-house teeth-photo database collected using the proposed application from 92 subjects each providing 10 samples in multiple sessions over a span of 3–4 days. It has been observed that the proposed system achieved 97.61% accuracy with a Correct Recognition Rate (CRR) of 95% at an Equal Error Rate (EER) as low as 2.07% even for a small RoI of size 175 × 175. To the best of our understanding, this is the first work on teeth-photo-based authentication for mobile devices. The database along with the code is being made public.
Enabling AI in Agriculture 4.0: A Blockchain-Based Mobile CrowdSensing Architecture
(Springer, 2024-04) Bhatia, Ashutosh; Tiwari, Kamlesh
Agriculture 4.0 relies on extensive data for predictive services, necessitating effective data collection. Mobile CrowdSensing (MCS), with its cost-effectiveness and scalability, addresses this need but faces centralization limitations. Blockchain-based frameworks have been proposed to mitigate these issues but often focus solely on data collection, lacking a comprehensive end-to-end architecture for smart agriculture. Recent literature has explored the integration of the Internet of Things (IoT), edge computing, fog computing, and cloud computing capabilities to establish centralized end-to-end architectures. Nonetheless, these architectures come with their own set of centralized limitations. In the context of contemporary technologies, the integration of blockchain and digital twin (DT) holds the potential to revolutionize the field of smart agriculture. This paper introduces a holistic end-to-end, layered, and service-oriented architecture for Agriculture 4.0, integrating mobile crowdsensing, blockchain, and DT. Unlike existing architectures, this approach aims to overcome centralization limitations, leveraging the strengths of emerging technologies. The proposed architecture extends current capabilities for more efficient and secure Agriculture 4.0 practices. We deploy the suggested architecture onto the Ethereum blockchain, demonstrating its practicality through the obtained results.
Enhanced lightweight quantum key distribution protocol for improved efficiency and security
(IEEE, 2025) Bhatia, Ashutosh; Bitragunta, Sainath; Tiwari, Kamlesh
Quantum Key Distribution (QKD) provides secure communication by leveraging quantum mechanics, with the BB84 protocol being one of its most widely adopted implementations. However, the classical post-processing steps in BB84, such as sifting, error correction, and key verification, often result in significant communication overhead, limiting its efficiency and scalability. In this work, we propose three key optimizations for BB84: (1) PRNG-based predetermined key bit positioning, which eliminates redundant bit exchanges during sifting, (2) hash-based subsequence comparison, enabling lightweight and efficient key verification, and (3) adaptive basis reconciliation, which minimizes the communication costs associated with basis matching. The proposed optimizations achieve a 50% reduction in communication overhead for large key sizes compared to traditional QKD protocols, as demonstrated through rigorous performance analysis. While the focus of this work is on the BB84 protocol, these optimizations are also directly applicable to a broader class of Discrete-Variable QKD (DV-QKD) protocols, such as six-state, B92, and E91, which share a fundamentally similar post-processing structure. This generality highlights the modularity and adaptability of the proposed methods across diverse QKD implementations. The proposed optimizations enhance post-processing efficiency and scalability, enabling practical deployment in bandwidth-limited environments like IoT networks, secure financial systems, and defense communications, thereby supporting broader adoption of quantum communication systems.
Enhancing Mobile Crowdsensing Security: A Proof of Stake-Based Publisher Selection Algorithm to Combat Sybil Attacks in Blockchain-Assisted MCS Systems
(Springer, 2024-04) Bhatia, Ashutosh; Tiwari, Kamlesh
In a blockchain-assisted Mobile CrowdSensing (MCS) System, individuals can generate as many blockchain identities as they desire, facilitating the execution of a Sybil attack. A Sybil attack can significantly impact such a system due to incorporating a reward mechanism and a majority-based data validation mechanism. An attacker can launch a Sybil attack with selfish or malicious intentions to maximize benefits from the system or to narrow down the reputation of the data requester (subscriber) and the system. Consequently, a Sybil attacker can discourage honest data collectors (publishers) and subscribers from participating, impeding the system’s potential success. In this paper, we propose a Sybil attack prevention cum avoidance mechanism to narrow down the effect of it in the blockchain-based MCS systems while maintaining the system’s requirements. The proposed mechanism incorporates a novel randomized publisher selection algorithm, leveraging the Proof-of-Stake (PoS) concept to render executing a Sybil attack costly and impractical. The simulation results show the effectiveness of the proposed mechanism.
Enhancing security through continuous biometric authentication using wearable sensors
(Elsevier, 2024) Bhatia, Ashutosh; Tiwari, Kamlesh
The paper presents a novel approach for biometric continuous driver authentication (CDA) for secure and safe transportation using wearable photoplethysmography (PPG) sensors and deep learning. Conventional one-time authentication (OTA) methods, while effective for initial identity verification, fail to continuously verify the driver’s identity during vehicle operation, potentially leading to safety, security, and accountability issues. To address this, we propose a system that employs Long Short-Term Memory (LSTM) models to predict subsequent PPG values from wrist-worn devices and continuously compare them with real-time sensor data for authentication. Our system calculates a confidence level representing the probability that the current user is the authorized driver, ensuring robust availability to genuine users while detecting impersonation attacks. The raw PPG data is directly fed into the LSTM model without pre-processing, ensuring lightweight processing. We validated our system with PPG data from 15 volunteers driving for 15 min in varied conditions. The system achieves an Equal Error Rate (EER) of 4.8%. Our results demonstrate that the system is a viable solution for CDA in dynamic environments, ensuring transparency, efficiency, accuracy, robust availability, and lightweight processing. Thus, our approach addresses the main challenges of classical driver authentication systems and effectively safeguards passengers and goods with robust driver authentication.
ExProCO: an explainable probabilistic campaign optimizer for ecommerce advertising
(Springer, 2025-04) Tiwari, Kamlesh; Bhatia, Ashutosh
Optimizing advertising (Ad) campaigns on e-commerce platforms is a complex task that extends beyond identifying correlations between budgets and bids or predicting metrics such as impression share and Return on Advertising Spend (ROAS). Effective Ad optimization involves addressing the auction-based nature of e-commerce advertising, which inherently requires a probabilistic approach to account for uncertainties and dynamic conditions. Consequently, a novel approach that combines probability analysis with decision tree modeling is proposed. The proposed model ExProCO is built on a Greedy-Modal Tree (GMT), offering a well-interpreted strategy for extracting useful information from complex, high-dimensional data at the campaign, keyword, and product levels. Using GMT, ExProCO minimizes Decision Tree instability and provides explainability. The Joint Probability Distribution (JPD) Model uses daily campaign data to explain how variables like bids and budget affect campaign outcomes. The ExProCO model outperforms other Machine Learning models in a benchmark comparison using financial data from a leading e-commerce firm. Achieving 75.4% accuracy with just 11.3 nodes, ExProCO excels in interpretability, noise resilience, and overfitting reduction, significantly improving campaign profitability and ad spend optimization.
Finger Knuckleprint Based Personal Authentication Using Siamese Network
(IEEE, 2019) Gupta, Karunesh Kumar; Tiwari, Kamlesh
Online security is a major concern today and incidents of forged identity cards and hacked passwords are common throughout the world. Therefore, there is a need for robust personal authentication mechanisms using biometrics for various access control systems. Popular biometric traits such as fingerprint have problems in rural areas, due to wearing down of fingerprint pattern from hard manual labor. This is also a problem for people who work with calcium oxide, because it is known to dissolve the upper layers of the skin due to its basicity. This paper proposes a finger-knuckle-print (FKP) based human authentication system that is immune to the above problems because the finger dorsal region is not exposed to labor surfaces. The paper uses pre-processed knuckle ROI images to train a Siamese convolutional neural network model. The proposed algorithm has been validated using open-source PolyU finger-knuckle-print database from 165 individuals, and has achieved 99.24% CRR, 0.78% EER that is better than the state-of-the-art.
A fractional-order model to study the dynamics of the spread of crime
(Elsevier, 2023-07) Mathur, Trilok; Tiwari, Kamlesh
Numerous crucial factors and parameters influence the dynamic process of the spread of crime. Various integer-order differential models have been proposed to capture crime spread. Most of these introduced dynamic systems have not considered the history of the criminal and the impact of crime on society. To address these shortcomings, a fractional-order crime transmission model is proposed in this manuscript considering five different classes law-abiding citizens, non-incarcerated criminals, incarcerated criminals, prison-released and recidivists. The primary focus of the proposed model is to study the effect of recidivism in society and decide the adequate imprisonment for repeat offenders. The existence, uniqueness, non-negativity and boundedness of the solution of the proposed model are examined. The local stability of the equilibrium points is also analysed using Routh–Hurwitz Criteria with Matignon conditions. Further, the threshold condition for the uniform asymptotic stability of the system is evaluated by the Lyapunov stability method. Moreover, the long-term impact of the imprisonment of criminals on society is also examined in the current study. The numerical simulations of the model for a range of fractional orders are obtained using power series expansion method to strengthen the theoretical results.
Hierarchical Classification using Neighbourhood Exploration for Sparse Text Tweets
(IEEE, 2022) Pasari, Sumanta; Tiwari, Kamlesh
Twitter has grown into a vast network of small informal text, and navigating it often becomes difficult for us. Here, we explore Natural Language Processing (NLP) approaches to make the topic classification of tweets easier. We do so with the use case for filtering non-profit tweets among different categories which are arranged in a hierarchy. This paper proposes an efficient pipeline for filtering relevant tweets and a novel data augmentation strategy for sparse datasets. Our data augmentation technique shows a significant leap in the training metrics and the accuracy on the test data increases by 9.52% and the F1-score by 24.82%.
I Know Who You are: A Learning Framework to Profile Smartphone Users
(IEEE, 2020) Bhatia, Ashutosh; Tiwari, Kamlesh
The volume of traffic belonging to mobile applications over the Internet has already crossed the traffic generated due to traditional desktop-based Internet browsing. To protect security and privacy of the mobile user, many smartphone applications use encryption to encapsulates communications over the Internet. However it is not possible to decode the actual message contents, even then the statistical information present in the traffic is useful to identify application and the associated activity. This paper proposes a machine learning based framework to analyze the encrypted mobile traffic with the objective of finding the mobile application usage patterns of smartphone users. This information can be used by various authorities to profile mobile users with respect to their age, gender profession, etc. which would further help them to look for any suspicious or anomalous behaviour. Proposed framework has been tested on the network traffic data pertaining to popular smartphone applications such as GMail, Facebook, and Youtube and have achieved 97.46% accuracy for application identification. Further, the proposed framework also achieved a fairly high accuracy, 77.37% when used for the classification of exact activity performed by the smartphone user in different applications
Is this URL Safe: Detection of Malicious URLs Using Global Vector for Word Representation
(IEEE, 2022) Bhatia, Ashutosh; Tiwari, Kamlesh
Users are frequently exposed to many unknown links through advertisements and emails. These links may contain URLs to mount targeted attacks like spamming, phishing, and malware installation. Using blacklist of URLs is the most widely used defense mechanism to detect a malicious URLs. However, automatically generating such a list for fresh malicious URLs is challenging. Detecting a URL as malicious using the lexicographical approach is an important research problem. This paper proposes a malicious URL detection mechanism using natural language processing. We use features including word vector representation obtained through GloVe along with statistical cues and n-gram on blacklist words. The proposed approach is efficient, and it does not require inputs from external servers to identify malicious URLs. Experiments are performed on 227,909 size database containing 80,128 benign and 147,781 malicious URLs. Proposed system has achieved an accuracy of 89% for ANN model with GloVe based features.
Layered blockchain-based mobile crowdsensing architecture: exploring privacy and scalability challenges across layers
(Springer, 2025-04) Bhatia, Ashutosh; Tiwari, Kamlesh
Blockchain technology has emerged as a transformative solution for addressing the limitations of traditional Mobile CrowdSensing (MCS) systems, which rely on centralized architectures. Despite its promise, the integration of blockchain into MCS introduces challenges related to privacy, scalability, and system efficiency. This paper presents a comprehensive layered architecture for enhancing blockchain-based MCS systems (BMCS), focusing on two critical dimensions: privacy and scalability. By categorizing challenges and proposed mitigation strategies, the study explores privacy risks arising from blockchain transparency and evaluates privacy-preserving mechanisms, including zero-knowledge proofs, multiparty computation, and homomorphic encryption, to protect sensitive data in decentralized environments. Scalability constraints, such as limited transaction throughput and resource intensity, are presented with targeted solutions that reduce on-chain loads and improve performance. The findings contribute actionable insights to advance BMCS systems, charting a path for resilient and scalable decentralized ecosystems.
A layered framework for blockchain security: classification of threats and the quantum computing impact
(Springer, 2025-04) Bhatia, Ashutosh; Tiwari, Kamlesh
Blockchain technology, with its transformative potential across industries, has ushered in a new era of decentralized systems. However, its widespread adoption has exposed vulnerabilities at various layers of its architecture, posing significant challenges to security and integrity. This paper introduces a comprehensive layered framework for blockchain security, classifying threats across five architectural layers: Application, Contract, Consensus, Network, and Data. By mapping vulnerabilities to these layers, the framework highlights specific attack vectors, such as Reentrancy, Sybil, Selfish Mining, and Replay attacks, and provides targeted mitigation strategies. Furthermore, the paper examines the disruptive potential of quantum computing on blockchain security, emphasizing the need for post-quantum cryptographic solutions to future-proof blockchain systems. The classification and analysis aim to guide researchers and developers in enhancing blockchain robustness. The findings contribute actionable insights into securing blockchain ecosystems and charting future research directions, including addressing interoperability challenges, optimizing smart contract security, and strengthening consensus mechanisms against evolving threats.