Sarita Yadav
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
Sarita Yadav has a total of 50 co-authors. Here are the top co-authors by number of shared publications:
- Achin Jain
Department of Information Technology, Bharati Vidyapeeth’s College of Engineering, New Delhi, India
2 shared publications - Arun Kumar Dubey
Department of Information Technology, Bharati Vidyapeeth’s College of Engineering, New Delhi, India
2 shared publications - Swaran Ahuja
Department of MDE, NorthCap University, Gurugram, India
2 shared publications Achin Jain
Bharati Vidyapeeth’s College of Engineering
1 shared publication- Alessandro Carriero
University of the Eastern Piedmont
1 shared publication - Alessio Paschè
Department of Radiology, Azienda Ospedaliero Universitaria (A.O.U.), 09124 Cagliari, Italy
1 shared publication
Latest Publications
- Open Access
- 13 October 2026
- Published by Springer Nature in Discover Artificial Intelligence
- Vol. 6 (1) , 1462
A lightweight explainable hybrid convolutional neural network for potato leaf blight detection
- 13 October 2026
- Published by Springer Nature in Discover Artificial Intelligence
- Vol. 6 (1) , 1462
Open AccessA lightweight explainable hybrid convolutional neural network for potato leaf blight detection
- 2 January 2026
- Published by Springer Nature
- p. 389-405
Study of Under Water Optical Communication System Under the Influence of Turbulence
- 2 January 2026
- Published by Springer Nature
- p. 389-405
Study of Under Water Optical Communication System Under the Influence of Turbulence
- 26 March 2024
- Published by Springer Nature
- p. 100-110
Vision-Based Human Activity Recognition Using CNN and LSTM Architecture
- 26 March 2024
- Published by Springer Nature
- p. 100-110
Vision-Based Human Activity Recognition Using CNN and LSTM Architecture
- 1 January 2024
- Published by Taru Publications in Journal of Information and Optimization Sciences
- Vol. 45 (2) , 533-543
A review on monitoring activities through human activity recognition
- 1 January 2024
- Published by Taru Publications in Journal of Information and Optimization Sciences
- Vol. 45 (2) , 533-543
A review on monitoring activities through human activity recognition
- Open Access
- 2 June 2023
- Published by MDPI AG in Diagnostics
- Vol. 13 (11) , 1954
Ensemble Deep Learning Derived from Transfer Learning for Classification of COVID-19 Patients on Hybrid Deep-Learning-Based Lung Segmentation: A Data Augmentation and Balancing Framework
Background and motivation: Lung computed tomography (CT) techniques are high-resolution and are well adopted in the intensive care unit (ICU) for COVID-19 disease control classification. Most artificial intelligence (AI) systems do not undergo generalization and are typically overfitted. Such trained AI systems are not practical for clinical settings and therefore do not give accurate results when executed on unseen data sets. We hypothesize that ensemble deep learning (EDL) is superior to deep transfer learning (TL) in both non-augmented and augmented frameworks. Methodology: The system consists of a cascade of quality control, ResNet–UNet-based hybrid deep learning for lung segmentation, and seven models using TL-based classification followed by five types of EDL’s. To prove our hypothesis, five different kinds of data combinations (DC) were designed using a combination of two multicenter cohorts—Croatia (80 COVID) and Italy (72 COVID and 30 controls)—leading to 12,000 CT slices. As part of generalization, the system was tested on unseen data and statistically tested for reliability/stability. Results: Using the K5 (80:20) cross-validation protocol on the balanced and augmented dataset, the five DC datasets improved TL mean accuracy by 3.32%, 6.56%, 12.96%, 47.1%, and 2.78%, respectively. The five EDL systems showed improvements in accuracy of 2.12%, 5.78%, 6.72%, 32.05%, and 2.40%, thus validating our hypothesis. All statistical tests proved positive for reliability and stability. Conclusion: EDL showed superior performance to TL systems for both (a) unbalanced and unaugmented and (b) balanced and augmented datasets for both (i) seen and (ii) unseen paradigms, validating both our hypotheses.
- 2 June 2023
- Published by MDPI AG in Diagnostics
- Vol. 13 (11) , 1954
Open AccessEnsemble Deep Learning Derived from Transfer Learning for Classification of COVID-19 Patients on Hybrid Deep-Learning-Based Lung Segmentation: A Data Augmentation and Balancing Framework
Background and motivation: Lung computed tomography (CT) techniques are high-resolution and are well adopted in the intensive care unit (ICU) for COVID-19 disease control classification. Most artificial intelligence (AI) systems do not undergo generalization and are typically overfitted. Such trained AI systems are not practical for clinical settings and therefore do not give accurate results when executed on unseen data sets. We hypothesize that ensemble deep learning (EDL) is superior to deep transfer learning (TL) in both non-augmented and augmented frameworks. Methodology: The system consists of a cascade of quality control, ResNet–UNet-based hybrid deep learning for lung segmentation, and seven models using TL-based classification followed by five types of EDL’s. To prove our hypothesis, five different kinds of data combinations (DC) were designed using a combination of two multicenter cohorts—Croatia (80 COVID) and Italy (72 COVID and 30 controls)—leading to 12,000 CT slices. As part of generalization, the system was tested on unseen data and statistically tested for reliability/stability. Results: Using the K5 (80:20) cross-validation protocol on the balanced and augmented dataset, the five DC datasets improved TL mean accuracy by 3.32%, 6.56%, 12.96%, 47.1%, and 2.78%, respectively. The five EDL systems showed improvements in accuracy of 2.12%, 5.78%, 6.72%, 32.05%, and 2.40%, thus validating our hypothesis. All statistical tests proved positive for reliability and stability. Conclusion: EDL showed superior performance to TL systems for both (a) unbalanced and unaugmented and (b) balanced and augmented datasets for both (i) seen and (ii) unseen paradigms, validating both our hypotheses.
- 7 October 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 70 (12) , 12651-12661
Performance Analysis and Altitude Optimization of UAV-Enabled Dual-Hop Mixed RF-UWOC System
This paper presents and investigates a dual-hop mixed radio frequency (RF) and underwater optical communication (UWOC) system. In the proposed system, an unmanned aerial vehicle (UAV) such as drone (S) located at low altitude is transmitting data signal towards the destination (D) which is located under water such as a submarine through a decode and forward (DF) relay (R) mounted on a ship over the sea surface. The data signal is transmitted from the UAV towards the relay located on the ship via RF channel modelled by the Nakagamim distributed fading statistics. The relay decodes the received RF data signal and re-transmits it towards the destination via visible light channel modelled by Exponential Generalized Gamma (EGG) distribution. Using the DF relaying protocol, we derive closed form analytical expressions for the outage probability and average bit error rate of the proposed mixed RF-UWOC system. Additionally, UAV optimal altitude analysis has been carried out to find the optimal elevation angle corresponding to the optimal altitude of the UAV where the system performance is maximized. The numerical simulation is carried out to support the performance analysis of the mixed RF-UWOC system and shows the influence of the various channel parameters such as air bubbles, UAV altitude, water salinity variations and scintillation on the end to end performance of the mixed RF-UWOC system.
- 7 October 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 70 (12) , 12651-12661
Performance Analysis and Altitude Optimization of UAV-Enabled Dual-Hop Mixed RF-UWOC System
This paper presents and investigates a dual-hop mixed radio frequency (RF) and underwater optical communication (UWOC) system. In the proposed system, an unmanned aerial vehicle (UAV) such as drone (S) located at low altitude is transmitting data signal towards the destination (D) which is located under water such as a submarine through a decode and forward (DF) relay (R) mounted on a ship over the sea surface. The data signal is transmitted from the UAV towards the relay located on the ship via RF channel modelled by the Nakagamim distributed fading statistics. The relay decodes the received RF data signal and re-transmits it towards the destination via visible light channel modelled by Exponential Generalized Gamma (EGG) distribution. Using the DF relaying protocol, we derive closed form analytical expressions for the outage probability and average bit error rate of the proposed mixed RF-UWOC system. Additionally, UAV optimal altitude analysis has been carried out to find the optimal elevation angle corresponding to the optimal altitude of the UAV where the system performance is maximized. The numerical simulation is carried out to support the performance analysis of the mixed RF-UWOC system and shows the influence of the various channel parameters such as air bubbles, UAV altitude, water salinity variations and scintillation on the end to end performance of the mixed RF-UWOC system.
- 17 February 2019
- Published by Taru Publications in Journal of Statistics and Management Systems
- Vol. 22 (2) , 379-391
Framework for SLEPS : Self learning electoral results prediction system
Computational social science is the study that focuses on developing and analyzing computation models for social issues. Elections are major social events that impact society. Behavioral analysis is a factor which affects electoral result prediction. It imbibes geographic comments, sentiment, and psychology of the participants. Deliberation in judgment based on gender, place of birth, native tongue and genealogy affects the results of electoral polls. Study of the impact of any event whether positive or negative that directly or indirectly influence the comments of the people stands as a factor affecting the elections. SLEPS is a proposed framework that uses large heterogeneous data set to train dual layer neural nets to increase the credibility and authenticity of polls. In SLEPS, we map data of any presidential or gubernatorial election comprising of social traits to statistical learning using neural nets. It is “self-learning’’ as it involves importance weighting of experts that incorporates features that affect election prediction viz. facial appearance, deliberate judgment, character traits and background data of the contestant. SLEPS was evaluated over Indian Elections of 1985 and predicted the winner with root mean square error of 0.04 and 76% precision.
- 17 February 2019
- Published by Taru Publications in Journal of Statistics and Management Systems
- Vol. 22 (2) , 379-391
Framework for SLEPS : Self learning electoral results prediction system
Computational social science is the study that focuses on developing and analyzing computation models for social issues. Elections are major social events that impact society. Behavioral analysis is a factor which affects electoral result prediction. It imbibes geographic comments, sentiment, and psychology of the participants. Deliberation in judgment based on gender, place of birth, native tongue and genealogy affects the results of electoral polls. Study of the impact of any event whether positive or negative that directly or indirectly influence the comments of the people stands as a factor affecting the elections. SLEPS is a proposed framework that uses large heterogeneous data set to train dual layer neural nets to increase the credibility and authenticity of polls. In SLEPS, we map data of any presidential or gubernatorial election comprising of social traits to statistical learning using neural nets. It is “self-learning’’ as it involves importance weighting of experts that incorporates features that affect election prediction viz. facial appearance, deliberate judgment, character traits and background data of the contestant. SLEPS was evaluated over Indian Elections of 1985 and predicted the winner with root mean square error of 0.04 and 76% precision.