Rishav Singh
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
Research Subjects
Rishav Singh has a total of 46 co-authors. Here are the top co-authors by number of shared publications:
- Sanjay Kumar Singh
Department of Computer Science and Engineering, Indian Institute of Technology (BHU) Varanasi, Varanasi, India
8 shared publications - Amit Kumar Singh
National Institute of Technology Patna
4 shared publications - Sandeep Sambhaji Udmale
Department of Computer Engineering and Information Technology, Veermata Jijabai Technological Institute (VJTI), Mumbai, India
4 shared publications - Abhinav Kumar
Department of Computer Science and Engineering, Indian Institute of Technology (BHU), Varanasi, India
3 shared publications - Ankit Vidyarthi
Department of CSE&IT, Jaypee Institute of Information Technology, Noida, India
3 shared publications Hari Om
Indian School of Mines
3 shared publications
Latest Publications
- 22 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 25 (21) , 39978-39985
Distilling Intelligence: Boosting Photonic Sensor Performance Through Knowledge Transfer
The evolution of fiber optic surface plasmon resonance (FO-SPR) sensors has significantly advanced the field of sensing, driving innovations in biomedical diagnostics, environmental monitoring, and industrial automation. However, traditional machine learning approaches used to enhance these sensors often suffer from high prediction errors, restricting their impact. In this work, we present a sensor-focused knowledge distillation framework aimed at improving the prediction error and efficiency of Figure of Merit (FOM) prediction, an essential parameter for FO-SPR sensor optimization. The proposed method leverages an ensemble of fully connected neural networks (FNNs) as teacher models to extract nonlinear relationships between structural parameters and sensor performance. These models generate soft labels that encapsulate intermodel uncertainty and latent patterns. A compact student model is then trained using a dual-loss function that integrates both soft teacher outputs and ground-truth data. The resulting distilled model outperforms existing methods (e.g., GPR), achieving high performance (MSE: 8223.022, RMSE: 90.681, MAE: 50.712, and R2: 0.98). Residual analysis validates its robustness under varying conditions. This approach enables real-time, data-driven optimization of photonic sensor design, particularly in selecting optimal wavelengths and material thicknesses, thus representing a scalable solution for next-generation intelligent sensor systems.
- 22 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 25 (21) , 39978-39985
Distilling Intelligence: Boosting Photonic Sensor Performance Through Knowledge Transfer
The evolution of fiber optic surface plasmon resonance (FO-SPR) sensors has significantly advanced the field of sensing, driving innovations in biomedical diagnostics, environmental monitoring, and industrial automation. However, traditional machine learning approaches used to enhance these sensors often suffer from high prediction errors, restricting their impact. In this work, we present a sensor-focused knowledge distillation framework aimed at improving the prediction error and efficiency of Figure of Merit (FOM) prediction, an essential parameter for FO-SPR sensor optimization. The proposed method leverages an ensemble of fully connected neural networks (FNNs) as teacher models to extract nonlinear relationships between structural parameters and sensor performance. These models generate soft labels that encapsulate intermodel uncertainty and latent patterns. A compact student model is then trained using a dual-loss function that integrates both soft teacher outputs and ground-truth data. The resulting distilled model outperforms existing methods (e.g., GPR), achieving high performance (MSE: 8223.022, RMSE: 90.681, MAE: 50.712, and R2: 0.98). Residual analysis validates its robustness under varying conditions. This approach enables real-time, data-driven optimization of photonic sensor design, particularly in selecting optimal wavelengths and material thicknesses, thus representing a scalable solution for next-generation intelligent sensor systems.
- 16 June 2025
- Published by Elsevier in Computers and Electrical Engineering
- Vol. 126
3DSPECSN: Adaptive 3D spatial patch based siamese network for robust hyperspectral image analysis
- 16 June 2025
- Published by Elsevier in Computers and Electrical Engineering
- Vol. 126
3DSPECSN: Adaptive 3D spatial patch based siamese network for robust hyperspectral image analysis
- 4 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Letters
- Vol. 9 (1) , 1-4
Enhanced Prediction and Optimization of Oil–Water Emulsion Stability Through Application of Machine Learning and Explainable Artificial Intelligence on TFBG Sensor Data
In petroleum industry, crude oil is extracted in the form of an oil–water emulsion (OWE). Understanding this emulsion is crucial for further processing and utilization. This research investigates the use of experimental data from a tilted fiber Bragg grating (TFBG) sensor to characterize and quantify the stability of OWE while focusing on the impact analysis of parameters using machine learning (ML) and explainable artificial intelligence (XAI) techniques. The dataset consisting of experimental TFBG spectra (wavelength range: 1250–1650 nm) included parameters such as revolutions per minute (RPM) of the rotator, surfactant concentration (C s ), and area measurements (indicating OWE stability). Applying interpolation techniques, the dataset was augmented for effective training and testing of ML models. The results indicated that the random forest model enabled the highest R 2 value of 99.2% on the test data. Then, XAI techniques (namely, shapley additive explanations and local interpretable model-agnostic explanations were applied (from both global and local interpretations viewpoints) to determine the contribution of each feature (i.e., RPM and C s ). It was found that C s has a significantly greater impact on OWE stability than RPM. Wavelength (λ) was subsequently included in the ML and XAI analyses. The results again confirmed that C s remained the most significant factor in determining OWE stability, with reasonably lesser impacts of λ and RPM. This study provides valuable insights into designing procedures and optimizing parameters for OWE stability during crude oil processing.
- 4 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Letters
- Vol. 9 (1) , 1-4
Enhanced Prediction and Optimization of Oil–Water Emulsion Stability Through Application of Machine Learning and Explainable Artificial Intelligence on TFBG Sensor Data
In petroleum industry, crude oil is extracted in the form of an oil–water emulsion (OWE). Understanding this emulsion is crucial for further processing and utilization. This research investigates the use of experimental data from a tilted fiber Bragg grating (TFBG) sensor to characterize and quantify the stability of OWE while focusing on the impact analysis of parameters using machine learning (ML) and explainable artificial intelligence (XAI) techniques. The dataset consisting of experimental TFBG spectra (wavelength range: 1250–1650 nm) included parameters such as revolutions per minute (RPM) of the rotator, surfactant concentration (C s ), and area measurements (indicating OWE stability). Applying interpolation techniques, the dataset was augmented for effective training and testing of ML models. The results indicated that the random forest model enabled the highest R 2 value of 99.2% on the test data. Then, XAI techniques (namely, shapley additive explanations and local interpretable model-agnostic explanations were applied (from both global and local interpretations viewpoints) to determine the contribution of each feature (i.e., RPM and C s ). It was found that C s has a significantly greater impact on OWE stability than RPM. Wavelength (λ) was subsequently included in the ML and XAI analyses. The results again confirmed that C s remained the most significant factor in determining OWE stability, with reasonably lesser impacts of λ and RPM. This study provides valuable insights into designing procedures and optimizing parameters for OWE stability during crude oil processing.
- RetractionOpen Access
- 2 December 2024
- Published by Springer Nature in Journal of Ambient Intelligence and Humanized Computing
- Vol. 15 (S1) , 257
Retraction Note: Identifying tiny faces in thermal images using transfer learning
- 2 December 2024
- Published by Springer Nature in Journal of Ambient Intelligence and Humanized Computing
- Vol. 15 (S1) , 257
RetractionOpen AccessRetraction Note: Identifying tiny faces in thermal images using transfer learning
- 22 April 2024
- Published by Elsevier in Optical Fiber Technology
- Vol. 85
On the application of explainable AI in optimizing the performance and design of fiber optic SPR sensor
- 22 April 2024
- Published by Elsevier in Optical Fiber Technology
- Vol. 85
On the application of explainable AI in optimizing the performance and design of fiber optic SPR sensor
- 5 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Fuzzy Systems
- Vol. 32 (10) , 5557-5570
SLIDE-Net: A Sequential Modeling Approach With Adaptive Fuzzy C-Mean Empowered Data Balancing Policy for IDC Detection
Breast cancer is a significant global health concern, with Invasive Ductal Carcinoma (IDC) being a significant subtype. Detecting IDC is a challenging task that can be hindered by the oversight of important contextual cues within Whole Slide Images (WSIs). To address this issue, we present the SequentialLSTM Invasive Ductal Carcinoma Detection with EfficientNet (SLIDE-Net) framework. SLIDE-Net synchronizes tissue image patch locations within WSIs, allowing for comprehensive and subtle identification of IDC. The inherent issue of data imbalance in IDC datasets, particularly the variable density of positive IDC patches concerning WSI size, is effectively tackled through the introduction of Adaptive Fuzzy C-Mean as a Data Balancing Policy. This novel approach enhances model efficacy and robustness, resulting in superior performance across key metrics such as accuracy (86%), balanced accuracy (86%), sensitivity (87%), specificity (86%), and GMean (86%). Our findings were substantiated by a comprehensive analysis revealing the significant cross-patch dependencies observed among patches of similar classes. When tested on the balanced PatchCAM dataset, SLIDENet once again showcased superiority with accuracy (89%), balanced accuracy (89%), F1-Score (87%), sensitivity (87%), and specificity (90%). This underscores the efficient utilization of shared contextual information within WSIs. These results firmly establish SLIDE-Net as a robust and reliable solution for accurate IDC detection. Our work not only advances the precision of IDC detection but also contributes valuable insights into the intricate dynamics of histopathology image analysis, paving the way for enhanced diagnostic accuracy in the ongoing battle against breast cancer. Index Terms—Breast Cancer, Sequential-Modeling, Data Balancing Policy, WSI, Adaptive Fuzzy C-Mean, Deep Learning.
- 5 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Fuzzy Systems
- Vol. 32 (10) , 5557-5570
SLIDE-Net: A Sequential Modeling Approach With Adaptive Fuzzy C-Mean Empowered Data Balancing Policy for IDC Detection
Breast cancer is a significant global health concern, with Invasive Ductal Carcinoma (IDC) being a significant subtype. Detecting IDC is a challenging task that can be hindered by the oversight of important contextual cues within Whole Slide Images (WSIs). To address this issue, we present the SequentialLSTM Invasive Ductal Carcinoma Detection with EfficientNet (SLIDE-Net) framework. SLIDE-Net synchronizes tissue image patch locations within WSIs, allowing for comprehensive and subtle identification of IDC. The inherent issue of data imbalance in IDC datasets, particularly the variable density of positive IDC patches concerning WSI size, is effectively tackled through the introduction of Adaptive Fuzzy C-Mean as a Data Balancing Policy. This novel approach enhances model efficacy and robustness, resulting in superior performance across key metrics such as accuracy (86%), balanced accuracy (86%), sensitivity (87%), specificity (86%), and GMean (86%). Our findings were substantiated by a comprehensive analysis revealing the significant cross-patch dependencies observed among patches of similar classes. When tested on the balanced PatchCAM dataset, SLIDENet once again showcased superiority with accuracy (89%), balanced accuracy (89%), F1-Score (87%), sensitivity (87%), and specificity (90%). This underscores the efficient utilization of shared contextual information within WSIs. These results firmly establish SLIDE-Net as a robust and reliable solution for accurate IDC detection. Our work not only advances the precision of IDC detection but also contributes valuable insights into the intricate dynamics of histopathology image analysis, paving the way for enhanced diagnostic accuracy in the ongoing battle against breast cancer. Index Terms—Breast Cancer, Sequential-Modeling, Data Balancing Policy, WSI, Adaptive Fuzzy C-Mean, Deep Learning.
- 12 February 2024
- Published by Springer Nature in Multimedia Tools and Applications
- Vol. 83 (33) , 80129-80159
Biometrics recognition of newborn: a review
- 12 February 2024
- Published by Springer Nature in Multimedia Tools and Applications
- Vol. 83 (33) , 80129-80159
Biometrics recognition of newborn: a review
- 21 July 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 23 (17) , 20330-20337
Exploring Deep Learning Models Aimed at Favorable Optimization and Enhancement of Fiber Optic Sensor’s Performance
We present a deep learning (DL)-assisted extrapolation approach for modeling a function that maximizes the performance of a fiber optic sensor (in terms of figure of merit, i.e ., FOM) under the variation of design parameters. We exploit a presumption that FOM should be treated as a relative entity. Recurrent Neural Network (RNN) framework is used to build the models that treat the data sequentially and relatively detect the hidden patterns. The objective of the adopted methodology is to forecast the best probable value of FOM for a particular combination of design parameters considered under finer resolutions. Our proposed model displays promising results on the test set and in the extrapolation corresponding to sub-sequence length ( I ) value of 35. Specifically, the Seq2Seq-Attnention model’s performance on the test metrics is RMSE = 2.12 ±0.62, MAE = 0.54±0.16, and SMAPE = 0.92±0.18. To verify the generality of our model for finer resolution data, we conducted tests for different values of wavelength. The results show that the model can capture the pattern and momentum of the FOM but marginally struggles to keep track of its magnitude for finer resolution data. However, the predictions for the training resolution are adequate. The processing time to predict FOM is nearly 1 second (68 ms/step), at least 100 times lesser than traditional simulation techniques. Our proposed network is relatively robust, converges quickly, has fewer parameters, and performs well on the test dataset. The analysis enables highly efficient and cost-effective optimization and enhancement of the fiber optic sensor’s performance.
- 21 July 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 23 (17) , 20330-20337
Exploring Deep Learning Models Aimed at Favorable Optimization and Enhancement of Fiber Optic Sensor’s Performance
We present a deep learning (DL)-assisted extrapolation approach for modeling a function that maximizes the performance of a fiber optic sensor (in terms of figure of merit, i.e ., FOM) under the variation of design parameters. We exploit a presumption that FOM should be treated as a relative entity. Recurrent Neural Network (RNN) framework is used to build the models that treat the data sequentially and relatively detect the hidden patterns. The objective of the adopted methodology is to forecast the best probable value of FOM for a particular combination of design parameters considered under finer resolutions. Our proposed model displays promising results on the test set and in the extrapolation corresponding to sub-sequence length ( I ) value of 35. Specifically, the Seq2Seq-Attnention model’s performance on the test metrics is RMSE = 2.12 ±0.62, MAE = 0.54±0.16, and SMAPE = 0.92±0.18. To verify the generality of our model for finer resolution data, we conducted tests for different values of wavelength. The results show that the model can capture the pattern and momentum of the FOM but marginally struggles to keep track of its magnitude for finer resolution data. However, the predictions for the training resolution are adequate. The processing time to predict FOM is nearly 1 second (68 ms/step), at least 100 times lesser than traditional simulation techniques. Our proposed network is relatively robust, converges quickly, has fewer parameters, and performs well on the test dataset. The analysis enables highly efficient and cost-effective optimization and enhancement of the fiber optic sensor’s performance.
- 26 June 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Informatics
- Vol. 20 (2) , 2253-2262
A Label-Efficient Semi Self-Supervised Learning Framework for IoT Devices in Industrial Process
The industrial sector has experienced a tremendous advancement in deep supervised learning due to its representation ability, but it comes with high computing and labeled data demands. Recently, the demand for intelligent IoT devices on assembly and disassembly lines has surged. This necessitates algorithms that can use data to make intelligent decisions and a framework that can enable multiple IoT devices to learn collaboratively. Further, huge image generation using IoTs also needs an efficient data annotation scheme for classification problems. WeCollab is one such framework in federated learning that significantly reduces human efforts in data annotation with breakthroughs in self-supervised learning. The proposed framework is generic and can be adapted to any specific image data generated by industrial robots involved in assembly and disassembly lines. Our method outperforms supervised learning by 25% and 20% on the CIFAR-10 and CINIC-10 datasets, respectively, for the labeling task. We generate pseudo labels for the unlabeled part of the data and train a model to achieve 30% better test accuracy on CIFAR-10 and 20% better test accuracy on the CINIC-10 dataset as compared to supervised learning. Extensive experiments unveil the effectiveness of the method and proposed combination of loss functions used by WeCollab.
- 26 June 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Informatics
- Vol. 20 (2) , 2253-2262
A Label-Efficient Semi Self-Supervised Learning Framework for IoT Devices in Industrial Process
The industrial sector has experienced a tremendous advancement in deep supervised learning due to its representation ability, but it comes with high computing and labeled data demands. Recently, the demand for intelligent IoT devices on assembly and disassembly lines has surged. This necessitates algorithms that can use data to make intelligent decisions and a framework that can enable multiple IoT devices to learn collaboratively. Further, huge image generation using IoTs also needs an efficient data annotation scheme for classification problems. WeCollab is one such framework in federated learning that significantly reduces human efforts in data annotation with breakthroughs in self-supervised learning. The proposed framework is generic and can be adapted to any specific image data generated by industrial robots involved in assembly and disassembly lines. Our method outperforms supervised learning by 25% and 20% on the CIFAR-10 and CINIC-10 datasets, respectively, for the labeling task. We generate pseudo labels for the unlabeled part of the data and train a model to achieve 30% better test accuracy on CIFAR-10 and 20% better test accuracy on the CINIC-10 dataset as compared to supervised learning. Extensive experiments unveil the effectiveness of the method and proposed combination of loss functions used by WeCollab.
- 6 December 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 23 (3) , 2320-2327
Enhancing the Performance of Photonic Sensor Using Machine-Learning Approach
This paper reports on the implementation of adequate machine leaning (ML) models on different datasets vis-a-vis fiber optic plasmonic sensor device. The variation of sensor’s figure of merit (FOM) with light wavelength (λ) and metal layer thickness (d m ) is considered as a starting point and accordingly, the appropriate ML model is chosen. The FOM datasets were found to be consistent with the Gaussian Process Regressor (GPR) model. The application of GPR with finer resolution (0.001 nm) of λ on the datasets led to enhanced magnitudes of sensors FOM. The dataset (459 points) having 9 different values of d m led to a predicted FOM of 6526.23 at λ = 1099.343 nm. Further, the dataset (714 points) having 13 different values of d m led to a predicted FOM value 6356.98 at λ = 1099.345 nm. These are promising results as far as the application of the sensor in biosensing is concerned. Further, the chosen model is found to be highly consistent with the data in terms of trend matching, and the values of other evaluation parameters (e.g., R 2 and Mean absolute error etc.) are found to be in considerably desirable ranges. This study clearly reveals that the selection of appropriate ML model and its implementation on various datasets can lead to more efficient finalization of the sensor design with enhanced sensing performance. This process is critical before actual experimental realization of the finalized sensor design.
- 6 December 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 23 (3) , 2320-2327
Enhancing the Performance of Photonic Sensor Using Machine-Learning Approach
This paper reports on the implementation of adequate machine leaning (ML) models on different datasets vis-a-vis fiber optic plasmonic sensor device. The variation of sensor’s figure of merit (FOM) with light wavelength (λ) and metal layer thickness (d m ) is considered as a starting point and accordingly, the appropriate ML model is chosen. The FOM datasets were found to be consistent with the Gaussian Process Regressor (GPR) model. The application of GPR with finer resolution (0.001 nm) of λ on the datasets led to enhanced magnitudes of sensors FOM. The dataset (459 points) having 9 different values of d m led to a predicted FOM of 6526.23 at λ = 1099.343 nm. Further, the dataset (714 points) having 13 different values of d m led to a predicted FOM value 6356.98 at λ = 1099.345 nm. These are promising results as far as the application of the sensor in biosensing is concerned. Further, the chosen model is found to be highly consistent with the data in terms of trend matching, and the values of other evaluation parameters (e.g., R 2 and Mean absolute error etc.) are found to be in considerably desirable ranges. This study clearly reveals that the selection of appropriate ML model and its implementation on various datasets can lead to more efficient finalization of the sensor design with enhanced sensing performance. This process is critical before actual experimental realization of the finalized sensor design.