Neelam Sinha
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Neelam Sinha has a total of 35 co-authors. Here are the top co-authors by number of shared publications:
- Chakka Sai Pradeep
International Institute of Information Technology
6 shared publications - Gopika Gopan K.
Wadhwani AI, New Delhi, India
4 shared publications Debanjali Bhattacharya
Networking and Communication, International Institute of Information Technology, Bangalore, India
3 shared publications- Hariharan Ramasangu
Research Division, Relecura Inc, Bangalore, Karnataka, India
3 shared publications Jitender Saini
Neuroimaging and interventional radiology, National Institute of Mental Health and Neuro Science, Bengaluru, India
3 shared publications- Siva Manohar Reddy Kesu
AIThinktank, Qualizeal India Private Ltd, Hyderabad, Telangana, India
3 shared publications
Latest Publications
- Open Access
- 17 December 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-9
Is the Spatial Structure as Important as Temporal Structure in EEG Signals?
The conventional assumption in electroencephalography (EEG) analysis is the importance of temporal structure. This study challenges that view by systematically investigating the impact of temporal disruption on classification performance across five diverse datasets using a transformer model. We evaluated four conditions: the original signal, a time-reversed signal, a randomly shuffled signal that preserved instantaneous spatial topography, and a randomly shuffled signal that destroyed temporal and spatial structure. Our results reveal a surprising resilience to temporal disruption. For most tasks, including imagery and cognitive load, scrambling the temporal order while preserving cross-channel relationships caused only a minor degradation in performance. The MCC on an imagery task dropped minimally from 0.72 to 0.71 and prediction entropy increased by just 9.4%. The minimal accuracy decrease and change in prediction confidence suggest that spatial patterns can be more informative than the temporal sequence. However, this principle is not universal; for the Alzheimer’s and Frontotemporal Dementia diagnostic task, this disruption caused performance to plummet from 0.98 to 0.47 MCC and prediction entropy to surge from 0.029 to 0.706, demonstrating that temporal dynamics are indispensable in this context. We conclude that the relative importance of spatial versus temporal information in EEG is highly context-dependent, a finding that questions the universal primacy of temporal order and opens new avenues for model design in Brain-Computer Interfaces.
- 17 December 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-9
Open AccessIs the Spatial Structure as Important as Temporal Structure in EEG Signals?
The conventional assumption in electroencephalography (EEG) analysis is the importance of temporal structure. This study challenges that view by systematically investigating the impact of temporal disruption on classification performance across five diverse datasets using a transformer model. We evaluated four conditions: the original signal, a time-reversed signal, a randomly shuffled signal that preserved instantaneous spatial topography, and a randomly shuffled signal that destroyed temporal and spatial structure. Our results reveal a surprising resilience to temporal disruption. For most tasks, including imagery and cognitive load, scrambling the temporal order while preserving cross-channel relationships caused only a minor degradation in performance. The MCC on an imagery task dropped minimally from 0.72 to 0.71 and prediction entropy increased by just 9.4%. The minimal accuracy decrease and change in prediction confidence suggest that spatial patterns can be more informative than the temporal sequence. However, this principle is not universal; for the Alzheimer’s and Frontotemporal Dementia diagnostic task, this disruption caused performance to plummet from 0.98 to 0.47 MCC and prediction entropy to surge from 0.029 to 0.706, demonstrating that temporal dynamics are indispensable in this context. We conclude that the relative importance of spatial versus temporal information in EEG is highly context-dependent, a finding that questions the universal primacy of temporal order and opens new avenues for model design in Brain-Computer Interfaces.
- Open Access
- 17 December 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-9
Alzheimer’s Disease Classification using Retinal OCT: TransNetOCT and Swin Transformer Models
- 17 December 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-9
Open AccessAlzheimer’s Disease Classification using Retinal OCT: TransNetOCT and Swin Transformer Models
- Open Access
- 13 December 2024
- Published by Association for Computing Machinery (ACM)
- p. 1-8
Dementia Disease Progression Analysis from 2D Cellular Automata Architecture
- 13 December 2024
- Published by Association for Computing Machinery (ACM)
- p. 1-8
Open AccessDementia Disease Progression Analysis from 2D Cellular Automata Architecture
- 4 December 2024
- Published by Springer Nature
- p. 332-344
Self-supervised Siamese Network Using Vision Transformer for Depth Estimation in Endoscopic Surgeries
- 4 December 2024
- Published by Springer Nature
- p. 332-344
Self-supervised Siamese Network Using Vision Transformer for Depth Estimation in Endoscopic Surgeries
- 24 July 2024
- Published by Springer Nature
- p. 529-536
Semi-supervised Learning Using Variational Autoencoder - A Cluster Based Approach
- 24 July 2024
- Published by Springer Nature
- p. 529-536
Semi-supervised Learning Using Variational Autoencoder - A Cluster Based Approach
- 24 July 2024
- Published by Springer Nature
- p. 151-159
Multi-stage Transfer Learning Based Yoga Pose Recognition Using CNN
- 24 July 2024
- Published by Springer Nature
- p. 151-159
Multi-stage Transfer Learning Based Yoga Pose Recognition Using CNN
- Open Access
- 11 July 2024
- Published by MDPI AG in Engineering Proceedings
- Vol. 68 (1) , 36
Cellular Automata Framework for Dementia Classification Using Explainable AI
The clinical dementia rating scale has been used for analyzing dementia severity based on cognitive impairments. Many researchers have introduced various statistical methods and machine learning techniques for the classification of dementia severity. Feature importance with deep learning architecture can give a better analysis of the dementia severity. A CA framework has been proposed for the classification of cognitive impairment, and LIME has been used for explaining the local interpretability. Feature vectors for healthy and unhealthy classes have been converted to redistributed CA images. These CA images have been classified using deep learning architecture, and promising results have been achieved. GRAD-CAM and LIME explainer have captured the feature importance of the cognitive impairment of CA images.
- 11 July 2024
- Published by MDPI AG in Engineering Proceedings
- Vol. 68 (1) , 36
Open AccessCellular Automata Framework for Dementia Classification Using Explainable AI
The clinical dementia rating scale has been used for analyzing dementia severity based on cognitive impairments. Many researchers have introduced various statistical methods and machine learning techniques for the classification of dementia severity. Feature importance with deep learning architecture can give a better analysis of the dementia severity. A CA framework has been proposed for the classification of cognitive impairment, and LIME has been used for explaining the local interpretability. Feature vectors for healthy and unhealthy classes have been converted to redistributed CA images. These CA images have been classified using deep learning architecture, and promising results have been achieved. GRAD-CAM and LIME explainer have captured the feature importance of the cognitive impairment of CA images.
- Open Access
Towards understanding the nature of direct functional connectivity in visual brain network
Recent advances in neuroimaging have enabled studies in functional connectivity (FC) of human brain, alongside investigation of the neuronal basis of cognition. One important FC study is the representation of vision in human brain. The release of publicly available dataset BOLD5000 has made it possible to study the brain dynamics during visual tasks in greater detail. In this paper, a comprehensive analysis of fMRI time series (TS) has been performed to explore different types of visual brain networks (VBN). The novelty of this work lies in (1) constructing VBN with consistently significant direct connectivity using both marginal and partial correlation, which is further analyzed using graph theoretic measures, (2) classification of VBNs as formed by image complexity-specific TS, using graphical features. In image complexity-specific VBN classification, XGBoost yields average accuracy in the range of 86.5% to 91.5% for positively correlated VBN, which is 2% greater than that using negative correlation. This result not only reflects the distinguishing graphical characteristics of each image complexity-specific VBN, but also highlights the importance of studying both positively correlated and negatively correlated VBN to understand the how differently brain functions while viewing different complexities of real-world images.
Open AccessTowards understanding the nature of direct functional connectivity in visual brain network
Recent advances in neuroimaging have enabled studies in functional connectivity (FC) of human brain, alongside investigation of the neuronal basis of cognition. One important FC study is the representation of vision in human brain. The release of publicly available dataset BOLD5000 has made it possible to study the brain dynamics during visual tasks in greater detail. In this paper, a comprehensive analysis of fMRI time series (TS) has been performed to explore different types of visual brain networks (VBN). The novelty of this work lies in (1) constructing VBN with consistently significant direct connectivity using both marginal and partial correlation, which is further analyzed using graph theoretic measures, (2) classification of VBNs as formed by image complexity-specific TS, using graphical features. In image complexity-specific VBN classification, XGBoost yields average accuracy in the range of 86.5% to 91.5% for positively correlated VBN, which is 2% greater than that using negative correlation. This result not only reflects the distinguishing graphical characteristics of each image complexity-specific VBN, but also highlights the importance of studying both positively correlated and negatively correlated VBN to understand the how differently brain functions while viewing different complexities of real-world images.
- Open Access
- 25 October 2023
- Published by Association for Computing Machinery (ACM)
- p. 1-6
YOLOv7-DocInstSeg: Efficient Instance Segmentation Framework for Document Analysis and Recognition
- 25 October 2023
- Published by Association for Computing Machinery (ACM)
- p. 1-6
Open AccessYOLOv7-DocInstSeg: Efficient Instance Segmentation Framework for Document Analysis and Recognition