Sakshi
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Sakshi has a total of 26 co-authors. Here are the top co-authors by number of shared publications:
- Vinay Kukreja
Chitkara University
10 shared publications - Chetan Sharma
Faculty of Management Studies, The ICFAI University, Baddi 174103, Himachal Pradesh, India
7 shared publications - Shamneesh Sharma
UpGrad Education Private Limited
4 shared publications - Ajit Noonia
Department of Computer Science and Engineering, Manipal University Jaipur, Jaipur, India
2 shared publications Chetan Sharma
UpGrad Campus, upGrad Education Private Limited,Bangalore,Karnataka,India
2 shared publications- Ajay Kumar
Manipal University Jaipur, Jaipur, India
1 shared publication
Latest Publications
- Open Access
- 14 November 2026
- Published by Springer Nature in Scientific Reports
Heuriskryptes framework for identifying covert channels in neural architectures
This paper introduces Heuriskryptes, a detection framework designed to identify AI-driven steganographic attacks within deep neural networks. The framework addresses the emerging threat of covert channels established within a model’s parameter pool. The evaluated attack vector, termed Stego-GCP, uses an orthogonal embedding to modify model weights with high stealth to leak data. To detect these channels, a specialized agent utilizes a Vision Transformer (ViT) architecture that treats the raw parameter space as an image, providing a self-attention-based mechanism that demonstrates consistent cross-architecture generalizability for efficient model auditing. The framework has been validated using a PooledCNN architecture on the Apple Disease and CIFAR-10 datasets. Results indicate that the carrier model retains a classification accuracy of 99.97%. Furthermore, the Heuriskryptes detector demonstrated high sensitivity, achieving 93.51% overall accuracy and maintaining a 90% detection rate specifically on the most challenging ambiguous, low-payload density scenarios ( $$\alpha < 0.002$$ ). This demonstrates highly competitive performance compared to evaluated frequency-domain anomaly detection baselines. Heuriskryptes serves as a reliable detection framework for verifying model integrity in collaborative and commercial AI environments. By focusing on the structural mathematical properties of the weight spectrum rather than activation-trigger patterns, this research provides a reliable defense against the unauthorized use of neural architectures as covert communication channels.- 14 November 2026
- Published by Springer Nature in Scientific Reports
Open AccessHeuriskryptes framework for identifying covert channels in neural architectures
This paper introduces Heuriskryptes, a detection framework designed to identify AI-driven steganographic attacks within deep neural networks. The framework addresses the emerging threat of covert channels established within a model’s parameter pool. The evaluated attack vector, termed Stego-GCP, uses an orthogonal embedding to modify model weights with high stealth to leak data. To detect these channels, a specialized agent utilizes a Vision Transformer (ViT) architecture that treats the raw parameter space as an image, providing a self-attention-based mechanism that demonstrates consistent cross-architecture generalizability for efficient model auditing. The framework has been validated using a PooledCNN architecture on the Apple Disease and CIFAR-10 datasets. Results indicate that the carrier model retains a classification accuracy of 99.97%. Furthermore, the Heuriskryptes detector demonstrated high sensitivity, achieving 93.51% overall accuracy and maintaining a 90% detection rate specifically on the most challenging ambiguous, low-payload density scenarios ( $$\alpha < 0.002$$ ). This demonstrates highly competitive performance compared to evaluated frequency-domain anomaly detection baselines. Heuriskryptes serves as a reliable detection framework for verifying model integrity in collaborative and commercial AI environments. By focusing on the structural mathematical properties of the weight spectrum rather than activation-trigger patterns, this research provides a reliable defense against the unauthorized use of neural architectures as covert communication channels. - 25 November 2025
- Published by Springer Nature
- p. 503-513
Integrative Framework for Accurate Gooseberry Leaf Disease Identification
- 25 November 2025
- Published by Springer Nature
- p. 503-513
Integrative Framework for Accurate Gooseberry Leaf Disease Identification
- Open Access
- 19 November 2024
- Published by Springer Nature in Discover Applied Sciences
- Vol. 6 (12) , 1-15
Advancements in handwritten Devanagari character recognition: a study on transfer learning and VGG16 algorithm
- 19 November 2024
- Published by Springer Nature in Discover Applied Sciences
- Vol. 6 (12) , 1-15
Open AccessAdvancements in handwritten Devanagari character recognition: a study on transfer learning and VGG16 algorithm
- 28 January 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 872-876
Data-Driven Approaches to Early Diabetes Prediction: A Comprehensive Exploration of Predictive Analytics and Machine Learning
Diabetes, a prevalent chronic condition globally, necessitates early prediction for timely interventions and improved patient outcomes. Leveraging data mining techniques and advancements in medical technology, this research explores the intersection of predictive analytics and medical interventions for early diabetes detection. The literature review highlights the effectiveness of data mining methods, including SVM Classification, and Logistic Regression, in predicting diabetes and estimating its severity. The proposed methodology involves data collection, preprocessing, and classification using algorithms like Naïve Bayes, BayesNet, SVM, Logistic Regression, and k-Nearest Neighbors. The dataset from the National Institute of Diabetes and Digestive and Kidney Diseases is subjected to rigorous preprocessing to ensure accurate model training. The classification algorithms are evaluated based on accuracy, precision, recall, and F-measure, with Naïve Bayes exhibiting the highest accuracy of 73.04%. Feature selection using the ranker algorithm emphasizes the significance of each attribute, contributing to the understanding of diabetes prediction. The study underscores the potential of data-driven approaches in healthcare, particularly for conditions like diabetes, and offers insights for future research and clinical applications.
- 28 January 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 872-876
Data-Driven Approaches to Early Diabetes Prediction: A Comprehensive Exploration of Predictive Analytics and Machine Learning
Diabetes, a prevalent chronic condition globally, necessitates early prediction for timely interventions and improved patient outcomes. Leveraging data mining techniques and advancements in medical technology, this research explores the intersection of predictive analytics and medical interventions for early diabetes detection. The literature review highlights the effectiveness of data mining methods, including SVM Classification, and Logistic Regression, in predicting diabetes and estimating its severity. The proposed methodology involves data collection, preprocessing, and classification using algorithms like Naïve Bayes, BayesNet, SVM, Logistic Regression, and k-Nearest Neighbors. The dataset from the National Institute of Diabetes and Digestive and Kidney Diseases is subjected to rigorous preprocessing to ensure accurate model training. The classification algorithms are evaluated based on accuracy, precision, recall, and F-measure, with Naïve Bayes exhibiting the highest accuracy of 73.04%. Feature selection using the ranker algorithm emphasizes the significance of each attribute, contributing to the understanding of diabetes prediction. The study underscores the potential of data-driven approaches in healthcare, particularly for conditions like diabetes, and offers insights for future research and clinical applications.
- 28 January 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 877-882
Green Intelligence: A Sequential CNN Odyssey in Mustard Leaf Disease Detection
Mustard plants are a crucial agricultural commodity for food and oil production. However, they are frequently vulnerable to various diseases that can substantially reduce crop yield. Early identification and diagnosis of these illnesses are essential for efficient management and control, ensuring sustainable production and agricultural output continuity. This research presents a new method for categorizing mustard leaf diseases using Convolutional Neural Networks (CNNs). The study utilizes sophisticated machine learning algorithms to differentiate between healthy and diseased mustard leaves, which is crucial for maintaining agricultural output and guaranteeing food security. The study involves training CNN architectures—Sequential CNN, ResNet-50, VGG, and AlexNet— on a meticulously curated dataset comprising images of healthy and diseased mustard leaves. The classifier's effectiveness is validated through comprehensive testing, demonstrating significant precision, recall, and F1-score advancements over conventional methods. This approach provides an efficient tool for disease detection in mustard crops and contributes to sustainable agricultural practices, aligning with the global goal of food security and environmental sustainability.
- 28 January 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 877-882
Green Intelligence: A Sequential CNN Odyssey in Mustard Leaf Disease Detection
Mustard plants are a crucial agricultural commodity for food and oil production. However, they are frequently vulnerable to various diseases that can substantially reduce crop yield. Early identification and diagnosis of these illnesses are essential for efficient management and control, ensuring sustainable production and agricultural output continuity. This research presents a new method for categorizing mustard leaf diseases using Convolutional Neural Networks (CNNs). The study utilizes sophisticated machine learning algorithms to differentiate between healthy and diseased mustard leaves, which is crucial for maintaining agricultural output and guaranteeing food security. The study involves training CNN architectures—Sequential CNN, ResNet-50, VGG, and AlexNet— on a meticulously curated dataset comprising images of healthy and diseased mustard leaves. The classifier's effectiveness is validated through comprehensive testing, demonstrating significant precision, recall, and F1-score advancements over conventional methods. This approach provides an efficient tool for disease detection in mustard crops and contributes to sustainable agricultural practices, aligning with the global goal of food security and environmental sustainability.
- 23 August 2023
- Published by Springer Nature in Multimedia Tools and Applications
- Vol. 83 (9) , 27831-27900
Machine learning and non-machine learning methods in mathematical recognition systems: Two decades’ systematic literature review
- 23 August 2023
- Published by Springer Nature in Multimedia Tools and Applications
- Vol. 83 (9) , 27831-27900
Machine learning and non-machine learning methods in mathematical recognition systems: Two decades’ systematic literature review
- 6 December 2022
- Published by Springer Nature in Artificial Intelligence Review
- Vol. 56 (7) , 7047-7135
A dive in white and grey shades of ML and non-ML literature: a multivocal analysis of mathematical expressions
- 6 December 2022
- Published by Springer Nature in Artificial Intelligence Review
- Vol. 56 (7) , 7047-7135
A dive in white and grey shades of ML and non-ML literature: a multivocal analysis of mathematical expressions
- 20 August 2022
- Published by Springer Nature in Multimedia Tools and Applications
- Vol. 81 (25) , 36805-36831
Latent DIRICHLET allocation (LDA) based information modelling on BLOCKCHAIN technology: a review of trends and research patterns used in integration
The past decade is known as the era of integrations where multiple technologies had integrated, and new research trends were seen. The security of data and information in the digital world has been a challenge to everyone; Blockchain technology has attracted many researchers in these scenarios. This paper focuses on finding the current trends in Blockchain technology to help the researchers select an area to carry future research. The data related to Blockchain Technologies have been collected from IEEE, Springer, ACM, and other digital databases. Then, the formulated corpus is used for topic modelling, and Latent Dirichlet Allocation is deployed. The outcomes of the Latent Dirichlet Allocation model are then analyzed based on various extracted key terms and key documents found for each topic. All the topic solution has been identified from the bag of words. The extracted topics are thereafter semantically mapped. Thus, based on the analysis of more than 900 papers, the most recent research trends have been discussed in this paper, ultimately focusing on the areas that need more attention from the research community. Also, the meta data analysis has been accomplished, evaluating the year wise and publication source wise research growth. More than 15 research directions are elaborated in this paper, which can direct and guide the researchers to pursuit the research in specific trends and also, find the research gaps in various technologies associated with Blockchain Technology.
- 20 August 2022
- Published by Springer Nature in Multimedia Tools and Applications
- Vol. 81 (25) , 36805-36831
Latent DIRICHLET allocation (LDA) based information modelling on BLOCKCHAIN technology: a review of trends and research patterns used in integration
The past decade is known as the era of integrations where multiple technologies had integrated, and new research trends were seen. The security of data and information in the digital world has been a challenge to everyone; Blockchain technology has attracted many researchers in these scenarios. This paper focuses on finding the current trends in Blockchain technology to help the researchers select an area to carry future research. The data related to Blockchain Technologies have been collected from IEEE, Springer, ACM, and other digital databases. Then, the formulated corpus is used for topic modelling, and Latent Dirichlet Allocation is deployed. The outcomes of the Latent Dirichlet Allocation model are then analyzed based on various extracted key terms and key documents found for each topic. All the topic solution has been identified from the bag of words. The extracted topics are thereafter semantically mapped. Thus, based on the analysis of more than 900 papers, the most recent research trends have been discussed in this paper, ultimately focusing on the areas that need more attention from the research community. Also, the meta data analysis has been accomplished, evaluating the year wise and publication source wise research growth. More than 15 research directions are elaborated in this paper, which can direct and guide the researchers to pursuit the research in specific trends and also, find the research gaps in various technologies associated with Blockchain Technology.