Shivani Agarwal
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Shivani Agarwal has a total of 26 co-authors. Here are the top co-authors by number of shared publications:
- Ayush Nanda
Odisha University of Technology and Research, India
2 shared publications - Aaditya Jain
Teerthanker Mahaveer University, Moradabad
1 shared publication Akanksha Shukla
Ajay Kumar Garg Engineering College, Ghaziabad, India
1 shared publication- Akanksha Shukla1 shared publication
- Anupama Sharma
Ajay Kumar Garg Engineering College, Ghaziabad
1 shared publication - Atul Kumar Rai
Kothiwal Institute of Technology and Professional Studies
1 shared publication
Latest Publications
- 5 August 2026
- Published by Taylor & Francis
- p. 121-139
Green Cloud Computing and Artificial Intelligence
In recent days, green cloud computing has gained a definite following despite the fact that this concept is just relatively new. This concept revolves around the usage of cloud computing in such a way that it has minimal impact on the environment. Strategies for this could include better usage of the cloud systems, development of more energy-efficient hardware and software, and even running cloud data centers on renewable energy. Such implementations of green cloud computing measures are also saving significant amounts of money to businesses and other enterprises, due to saving energies and carbon emission. It will help in reducing the impact of the cloud environment on the surroundings and fight global climate change. In this regard, the discussion topic is about green cloud computing, and this chapter outlines and discusses the numerous impacts, challenges, and financial implications of green cloud computing. Together with this, a discussion on how to apply the theory and practice will be done in business operations.
- 5 August 2026
- Published by Taylor & Francis
- p. 121-139
Green Cloud Computing and Artificial Intelligence
In recent days, green cloud computing has gained a definite following despite the fact that this concept is just relatively new. This concept revolves around the usage of cloud computing in such a way that it has minimal impact on the environment. Strategies for this could include better usage of the cloud systems, development of more energy-efficient hardware and software, and even running cloud data centers on renewable energy. Such implementations of green cloud computing measures are also saving significant amounts of money to businesses and other enterprises, due to saving energies and carbon emission. It will help in reducing the impact of the cloud environment on the surroundings and fight global climate change. In this regard, the discussion topic is about green cloud computing, and this chapter outlines and discusses the numerous impacts, challenges, and financial implications of green cloud computing. Together with this, a discussion on how to apply the theory and practice will be done in business operations.
- 5 August 2026
- Published by Taylor & Francis
- p. 200-217
MediQuery
With the fast-changing nature of healthcare, it is important that both medical professionals and patients have access to reliable, real-time information. But current artificial intelligence (AI)-based systems tend to lack the constraints of incomplete or unverifiable responses, leading to undesirable consequences, particularly in medicine. To mitigate this limitation, we introduce MediQuery, a Retrieval-Augmented Generation (RAG) model specifically designed for medical use. Medi Query uses Qdrant, a self-hosted vector database, BioMistral 7B, a modern medical language model, and PubMedBERT embeddings to enable precise, domain-specific information retrieval. Based on real-time data, the system is designed to provide exact, quick responses to challenging medical problems. Medi Query allows flawless performance and scalability by using FastAPI for the backend and LangChain for orchestration. Our approach increases the accuracy and dependability of medical query responses without sacrificing typical concerns including AI hallucinations. Medi Query is a strong competitor for next healthcare use cases since the results reveal significant performance Key Performance Indicator (KPIs) improvement. Future research will maximize scalability, solve privacy and bias concerns, and improve the model for certain medical fields.
- 5 August 2026
- Published by Taylor & Francis
- p. 200-217
MediQuery
With the fast-changing nature of healthcare, it is important that both medical professionals and patients have access to reliable, real-time information. But current artificial intelligence (AI)-based systems tend to lack the constraints of incomplete or unverifiable responses, leading to undesirable consequences, particularly in medicine. To mitigate this limitation, we introduce MediQuery, a Retrieval-Augmented Generation (RAG) model specifically designed for medical use. Medi Query uses Qdrant, a self-hosted vector database, BioMistral 7B, a modern medical language model, and PubMedBERT embeddings to enable precise, domain-specific information retrieval. Based on real-time data, the system is designed to provide exact, quick responses to challenging medical problems. Medi Query allows flawless performance and scalability by using FastAPI for the backend and LangChain for orchestration. Our approach increases the accuracy and dependability of medical query responses without sacrificing typical concerns including AI hallucinations. Medi Query is a strong competitor for next healthcare use cases since the results reveal significant performance Key Performance Indicator (KPIs) improvement. Future research will maximize scalability, solve privacy and bias concerns, and improve the model for certain medical fields.
- Open Access
- 1 April 2026
- Published by The ScienceIn Publishing in Journal of Integrated Science and Technology
- Vol. 14 (5) , 1569
SECDL-SEGNET: A secure Deep Learning segmentation network for intrusion detection in encrypted cyber traffic
Intrusion detection on encrypted network traffic has already proven to be more difficult with the popularity of encryption protocols like TLS and HTTPS that limit the inspection of packet payloads. Although encryption grants privacy and security of data, it equally facilitates criminal activities to be hidden in the normal flow of communication. The current intrusion detectors have a high dependency on a payload-based analysis or coarse-grained statistics and therefore cannot be utilized effectively in encrypted settings. In order to overcome these weaknesses, the present paper suggests a privacy-sensitive deep learning segmentation architecture, called SecDL-SegNet, to conduct intrusion detection in encrypted traffic. The given model uses flow-level statistical properties, metadata associated with packets, and temporal patterns of communication, without the need to decipher the payloads. An architecture that is based on segmentation is presented to divide the traffic into temporal windows so that a fine-grained analysis and localization of anomalies can be performed. The framework combines convolutional encoding, temporal modeling and computing segment-level anomaly score to improve the detection accuracy and interpretability. The proposed approach is proven to be effective as extensive experiments on the encrypted traffic benchmark datasets have been done. SecDL-SegNet has an average accuracy of 97.17, precision of 96.34, recall of 96.68 and F1-score of 96.59, which is stronger than the traditional machine learning and baseline deep learning predictors. Also, the model has low false positive of 2.18% and can be run in close real-time, with an average inference time of 5.32 ms per flow. The findings indicate that SecDL-SegNet developed a scalable, powerful, and protection conscious network intrusion detection system that works in the next generation encrypted networks.
- 1 April 2026
- Published by The ScienceIn Publishing in Journal of Integrated Science and Technology
- Vol. 14 (5) , 1569
Open AccessSECDL-SEGNET: A secure Deep Learning segmentation network for intrusion detection in encrypted cyber traffic
Intrusion detection on encrypted network traffic has already proven to be more difficult with the popularity of encryption protocols like TLS and HTTPS that limit the inspection of packet payloads. Although encryption grants privacy and security of data, it equally facilitates criminal activities to be hidden in the normal flow of communication. The current intrusion detectors have a high dependency on a payload-based analysis or coarse-grained statistics and therefore cannot be utilized effectively in encrypted settings. In order to overcome these weaknesses, the present paper suggests a privacy-sensitive deep learning segmentation architecture, called SecDL-SegNet, to conduct intrusion detection in encrypted traffic. The given model uses flow-level statistical properties, metadata associated with packets, and temporal patterns of communication, without the need to decipher the payloads. An architecture that is based on segmentation is presented to divide the traffic into temporal windows so that a fine-grained analysis and localization of anomalies can be performed. The framework combines convolutional encoding, temporal modeling and computing segment-level anomaly score to improve the detection accuracy and interpretability. The proposed approach is proven to be effective as extensive experiments on the encrypted traffic benchmark datasets have been done. SecDL-SegNet has an average accuracy of 97.17, precision of 96.34, recall of 96.68 and F1-score of 96.59, which is stronger than the traditional machine learning and baseline deep learning predictors. Also, the model has low false positive of 2.18% and can be run in close real-time, with an average inference time of 5.32 ms per flow. The findings indicate that SecDL-SegNet developed a scalable, powerful, and protection conscious network intrusion detection system that works in the next generation encrypted networks.
- 16 November 2025
- Published by Institute of Central Computation and Knowledge Inc. in ICCK Transactions on Machine Intelligence
- Vol. 2 (1) , 1-11
Secure and Decentralized Heart Sound Analysis using Federated Learning and Blockchain Technology
Early diagnosis of cardiac abnormalities depends on accurate classification of heart sounds, but centralized training methods run the danger of violating patient privacy. We thus propose a privacy-preserving and reliable heart sound abnormality detection system combining Blockchain Technology with Federated Learning (FL). Training is spread among seven clients, each simulating an independent data source, using a preprocessed dataset from the PhysioNet Challenge 2016 to enable distributed learning without sharing raw data. CNN-LSTM model using FedAvg achieved the best performance: 94\% accuracy, 0.90 precision, 0.96 recall, and an AUC of 0.98 among five deep learning architectures evaluated with FedAvg and FedProx strategies. Along with metadata including client ID and round number, SHA-256 hashes of local and global model weights were recorded on a local Ethereum blockchain following every communication round to guarantee model integrity. The hash of the final model is revalidated against the blockchain to confirm authenticity prior to deployment. It then guarantees safe, distributed, clinically valuable AI-based diagnostics by real-time classification of heart sounds as normal or abnormal.
- 16 November 2025
- Published by Institute of Central Computation and Knowledge Inc. in ICCK Transactions on Machine Intelligence
- Vol. 2 (1) , 1-11
Secure and Decentralized Heart Sound Analysis using Federated Learning and Blockchain Technology
Early diagnosis of cardiac abnormalities depends on accurate classification of heart sounds, but centralized training methods run the danger of violating patient privacy. We thus propose a privacy-preserving and reliable heart sound abnormality detection system combining Blockchain Technology with Federated Learning (FL). Training is spread among seven clients, each simulating an independent data source, using a preprocessed dataset from the PhysioNet Challenge 2016 to enable distributed learning without sharing raw data. CNN-LSTM model using FedAvg achieved the best performance: 94\% accuracy, 0.90 precision, 0.96 recall, and an AUC of 0.98 among five deep learning architectures evaluated with FedAvg and FedProx strategies. Along with metadata including client ID and round number, SHA-256 hashes of local and global model weights were recorded on a local Ethereum blockchain following every communication round to guarantee model integrity. The hash of the final model is revalidated against the blockchain to confirm authenticity prior to deployment. It then guarantees safe, distributed, clinically valuable AI-based diagnostics by real-time classification of heart sounds as normal or abnormal.
- 18 April 2025
- Published by IGI Global Scientific Publishing
- p. 175-202
Cloud Computing for Machine Learning and Cognitive Application
Cloud computing is becoming essential for the large-scale deployment of cognitive and machine learning (ML) applications. This chapter looks at how cloud platforms, which provide flexibility, cost-effectiveness, and on-demand resources, make it easier to utilize ML and cognitive technologies. An outline of key cloud services, such as data management, ML model building, and design patterns for effective processing, are among the important topics. We examine cognitive services, including computer vision and natural language processing, and show how cloud-based machine learning pipelines manage data deployment, training, and ingestion. The basis for safe and dependable cloud machine learning operations is provided by topics related to edge computing, security, and compliance. To improve operational flow, advanced techniques like MLOps are also covered. The chapter concludes with insights into new developments, such as cloud AI ethics and quantum computing, giving readers a road map for creating intelligent and scalable cloud applications.
- 18 April 2025
- Published by IGI Global Scientific Publishing
- p. 175-202
Cloud Computing for Machine Learning and Cognitive Application
Cloud computing is becoming essential for the large-scale deployment of cognitive and machine learning (ML) applications. This chapter looks at how cloud platforms, which provide flexibility, cost-effectiveness, and on-demand resources, make it easier to utilize ML and cognitive technologies. An outline of key cloud services, such as data management, ML model building, and design patterns for effective processing, are among the important topics. We examine cognitive services, including computer vision and natural language processing, and show how cloud-based machine learning pipelines manage data deployment, training, and ingestion. The basis for safe and dependable cloud machine learning operations is provided by topics related to edge computing, security, and compliance. To improve operational flow, advanced techniques like MLOps are also covered. The chapter concludes with insights into new developments, such as cloud AI ethics and quantum computing, giving readers a road map for creating intelligent and scalable cloud applications.
- Open Access
- 1 January 2025
- Published by National Library of Serbia in Facta universitatis - series: Electronics and Energetics
- Vol. 38 (3) , 431-456
Hybrid ai model for predicting air quality and rice crop yield using satellite data and environmental variables
Air quality has a significant impact on agricultural productivity, with pollutants such as ozone (O3), particulate matter (PM), and carbon monoxide (CO) posing serious threats to crop yields. Plant growth and yields can be disrupted by these pollutants, highlighting the need for effective solutions. Strategies, including improvements in air quality management to reduce pollutant levels and the development of advanced predictive models, have been proposed. The concentrations of air pollutants and their potential impacts on agriculture can be forecasted by these models, allowing proactive measures to be taken to mitigate adverse effects on crop yields. In this research, the challenges associated with predicting the impact of key air pollutants on rice yield in India are addressed. Satellite data from the Giovanni Data Centre was utilized to monitor concentrations of O?, PM, and CO, and to calculate the Air Quality Index (AQI). The Prophet model is employed to predict future pollutant levels and AQI. Soil temperature and air moisture data were incorporated to assess their combined impact on crop yield. Nineteen years of monthly rice yield data from FAOSTAT was used to train a feed-forward neural network with inputs including PM, O?, CO, AQI, soil temperature, and air moisture. A high accuracy of 94% was achieved by the model, effectively predicting crop yields based on these factors, and a clear inverse relationship between air quality and crop yield was demonstrated: significant decreases in yield were correlated with higher concentrations of O?, PM, and CO. Concrete evidence of the detrimental effects of air pollution on crop productivity is provided by these findings.
- 1 January 2025
- Published by National Library of Serbia in Facta universitatis - series: Electronics and Energetics
- Vol. 38 (3) , 431-456
Open AccessHybrid ai model for predicting air quality and rice crop yield using satellite data and environmental variables
Air quality has a significant impact on agricultural productivity, with pollutants such as ozone (O3), particulate matter (PM), and carbon monoxide (CO) posing serious threats to crop yields. Plant growth and yields can be disrupted by these pollutants, highlighting the need for effective solutions. Strategies, including improvements in air quality management to reduce pollutant levels and the development of advanced predictive models, have been proposed. The concentrations of air pollutants and their potential impacts on agriculture can be forecasted by these models, allowing proactive measures to be taken to mitigate adverse effects on crop yields. In this research, the challenges associated with predicting the impact of key air pollutants on rice yield in India are addressed. Satellite data from the Giovanni Data Centre was utilized to monitor concentrations of O?, PM, and CO, and to calculate the Air Quality Index (AQI). The Prophet model is employed to predict future pollutant levels and AQI. Soil temperature and air moisture data were incorporated to assess their combined impact on crop yield. Nineteen years of monthly rice yield data from FAOSTAT was used to train a feed-forward neural network with inputs including PM, O?, CO, AQI, soil temperature, and air moisture. A high accuracy of 94% was achieved by the model, effectively predicting crop yields based on these factors, and a clear inverse relationship between air quality and crop yield was demonstrated: significant decreases in yield were correlated with higher concentrations of O?, PM, and CO. Concrete evidence of the detrimental effects of air pollution on crop productivity is provided by these findings.
- Open Access
- 1 July 2023
- Published by Institute of Advanced Engineering and Science in International Journal of Reconfigurable and Embedded Systems (IJRES)
- Vol. 12 (2) , 240-247
Emotion classification for musical data using deep learning techniques
This research is done based on the identification and thorough analyzing musical data that is extracted by the various method. This extracted information can be utilized in the deep learning algorithm to identify the emotion, based on the hidden features of the dataset. Deep learning-based convolutional neural network (CNN) and long short-term memory-gated recurrent unit (LSTM-GRU) models were developed to predict the information from the musical information. The musical dataset is extracted using the fast Fourier transform (FFT) models. The three deep learning models were developed in this work the first model was based on the information of extracted information such as zero-crossing rate, and spectral roll-off. Another model was developed on the information of Mel frequencybased cepstral coefficient (MFCC) features, the deep and wide CNN algorithm with LSTM-GRU bidirectional model was developed. The third model was developed on the extracted information from Mel-spectrographs and untied these graphs based on two-dimensional (2D) data information to the 2D CNN model alongside LSTM models. Proposed model performance on the information from Mel-spectrographs is compared on the F1 score, precision, and classification report of the models. Which shows better accuracy with improved F1 and recall values as compared with existing approaches.
- 1 July 2023
- Published by Institute of Advanced Engineering and Science in International Journal of Reconfigurable and Embedded Systems (IJRES)
- Vol. 12 (2) , 240-247
Open AccessEmotion classification for musical data using deep learning techniques
This research is done based on the identification and thorough analyzing musical data that is extracted by the various method. This extracted information can be utilized in the deep learning algorithm to identify the emotion, based on the hidden features of the dataset. Deep learning-based convolutional neural network (CNN) and long short-term memory-gated recurrent unit (LSTM-GRU) models were developed to predict the information from the musical information. The musical dataset is extracted using the fast Fourier transform (FFT) models. The three deep learning models were developed in this work the first model was based on the information of extracted information such as zero-crossing rate, and spectral roll-off. Another model was developed on the information of Mel frequencybased cepstral coefficient (MFCC) features, the deep and wide CNN algorithm with LSTM-GRU bidirectional model was developed. The third model was developed on the extracted information from Mel-spectrographs and untied these graphs based on two-dimensional (2D) data information to the 2D CNN model alongside LSTM models. Proposed model performance on the information from Mel-spectrographs is compared on the F1 score, precision, and classification report of the models. Which shows better accuracy with improved F1 and recall values as compared with existing approaches.