Vibha Bhandari
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
Vibha Bhandari has a total of 2 co-authors. Here are the top co-authors by number of shared publications:
- Ghanahshyam B. Kshirsagar
College of Engineering, The Roux Institute, Northeastern University, Portland, ME, USA
4 shared publications - Narendra Digambar Londhe
Electrical Engineering Department, National Institute of Technology Raipur
4 shared publications
Latest Publications
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Human-Machine Systems
- Vol. 56 (1) , 124-134
P3-FSLNet: A Compact Spatio-Temporal Model With Contrastive Few-Shot Learning for Subject-Independent P300 Detection in Devanagari Script-Based P300 Speller
Brain–computer interface (BCI) systems frequently necessitate time-intensive subject-specific calibration, thereby motivating the development of subject-independent P300 detection approaches. Existing methodologies that employ transfer learning and knowledge distillation encounter challenges with limited generalizability due to substantial intersubject variability and constrained data availability. Furthermore, their applicability is often questioned, as most lack comprehensive external validation and cross-script evaluation. To address these limitations, we introduce P3-FSLNet, a few-shot metalearning framework that integrates prototypical networks and contrastive learning within a spatial-temporal convolutional neural network augmented by dual-channel attention for the selection of relevant electroencephalogram channels. Episodic metatraining facilitates the transfer of robust knowledge across different subjects. Evaluated on a self-recorded Devanagari script dataset, P3-FSLNet attains a classification accuracy of 93.17%, surpassing state-of-the-art methods by 1%–14%, while simultaneously reducing trainable parameters by up to 400 times. External validation using English-script datasets from BCI Competition II and III confirms its robustness in cross-subject and cross-script generalization. These findings demonstrate the efficacy of P3-FSLNet and represent a substantial advancement toward the development of script-agnostic P300 spellers that are lightweight, scalable, and conducive to multilingual applications.
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Human-Machine Systems
- Vol. 56 (1) , 124-134
P3-FSLNet: A Compact Spatio-Temporal Model With Contrastive Few-Shot Learning for Subject-Independent P300 Detection in Devanagari Script-Based P300 Speller
Brain–computer interface (BCI) systems frequently necessitate time-intensive subject-specific calibration, thereby motivating the development of subject-independent P300 detection approaches. Existing methodologies that employ transfer learning and knowledge distillation encounter challenges with limited generalizability due to substantial intersubject variability and constrained data availability. Furthermore, their applicability is often questioned, as most lack comprehensive external validation and cross-script evaluation. To address these limitations, we introduce P3-FSLNet, a few-shot metalearning framework that integrates prototypical networks and contrastive learning within a spatial-temporal convolutional neural network augmented by dual-channel attention for the selection of relevant electroencephalogram channels. Episodic metatraining facilitates the transfer of robust knowledge across different subjects. Evaluated on a self-recorded Devanagari script dataset, P3-FSLNet attains a classification accuracy of 93.17%, surpassing state-of-the-art methods by 1%–14%, while simultaneously reducing trainable parameters by up to 400 times. External validation using English-script datasets from BCI Competition II and III confirms its robustness in cross-subject and cross-script generalization. These findings demonstrate the efficacy of P3-FSLNet and represent a substantial advancement toward the development of script-agnostic P300 spellers that are lightweight, scalable, and conducive to multilingual applications.
- Open Access
- 5 March 2024
- Published by BON VIEW PUBLISHING PTE in Artificial Intelligence and Applications
- Vol. 2 (3) , 155-164
A Systematic Review of Computational Intelligence Techniques for Channel Selection in P300-Based Brain Computer Interface Speller
Electroencephalography (EEG)-based P300 speller aids in restoring the communication and control capabilities in patients suffering from motor disabilities. However, the quality and quantity of the data collected from EEG recordings have a substantial influence on the P300 speller’s performance. Hence, selecting the optimum number of recording electrodes, i.e., channels for each user, is a significant difficulty for the P300 speller. There are two fundamental objectives of the channel selection process: (1) to extract the most crucial information from the relevant channels, hence reducing the computing complexity of P300/non-P300 signal processing operation, and (2) to lessen the potential overfitting that could result from using unwanted channels to boost performance. For obtaining the best channel subsets, different channel selection techniques, including manual, filtering, wrapper, and embedded approaches, have been applied by past researchers. This research provides an in-depth examination of recent advancements, status, challenges, and potential solutions related to channel selection strategies in P300 speller systems. Each channel selection technique is thoroughly explored, including detailed comparisons between them. The notable advantages and drawbacks of each method are emphasized along with the discussion on the future direction and scope of work in the field of channel selection in P300 speller. The review underscores that channel selection methods enable the use of a reduced number of channels without compromising classification performance. By eliminating noisy or irrelevant channels, these approaches contribute to enhanced system performance. Received: 21 July 2023 | Revised: 3 January 2024 | Accepted: 29 February 2024 Conflicts of Interest Narendra D. Londhe is an Editorial Board Member for Artificial Intelligence and Applications, and was not involved in the editorial review or the decision to publish this article. The authors declare that they have no conflicts of interest to this work. Data Availability Statement Data sharing is not applicable to this article as no new data were created or analyzed in this study.
- 5 March 2024
- Published by BON VIEW PUBLISHING PTE in Artificial Intelligence and Applications
- Vol. 2 (3) , 155-164
Open AccessA Systematic Review of Computational Intelligence Techniques for Channel Selection in P300-Based Brain Computer Interface Speller
Electroencephalography (EEG)-based P300 speller aids in restoring the communication and control capabilities in patients suffering from motor disabilities. However, the quality and quantity of the data collected from EEG recordings have a substantial influence on the P300 speller’s performance. Hence, selecting the optimum number of recording electrodes, i.e., channels for each user, is a significant difficulty for the P300 speller. There are two fundamental objectives of the channel selection process: (1) to extract the most crucial information from the relevant channels, hence reducing the computing complexity of P300/non-P300 signal processing operation, and (2) to lessen the potential overfitting that could result from using unwanted channels to boost performance. For obtaining the best channel subsets, different channel selection techniques, including manual, filtering, wrapper, and embedded approaches, have been applied by past researchers. This research provides an in-depth examination of recent advancements, status, challenges, and potential solutions related to channel selection strategies in P300 speller systems. Each channel selection technique is thoroughly explored, including detailed comparisons between them. The notable advantages and drawbacks of each method are emphasized along with the discussion on the future direction and scope of work in the field of channel selection in P300 speller. The review underscores that channel selection methods enable the use of a reduced number of channels without compromising classification performance. By eliminating noisy or irrelevant channels, these approaches contribute to enhanced system performance. Received: 21 July 2023 | Revised: 3 January 2024 | Accepted: 29 February 2024 Conflicts of Interest Narendra D. Londhe is an Editorial Board Member for Artificial Intelligence and Applications, and was not involved in the editorial review or the decision to publish this article. The authors declare that they have no conflicts of interest to this work. Data Availability Statement Data sharing is not applicable to this article as no new data were created or analyzed in this study.
- Open Access
- 2 January 2024
- Published by BON VIEW PUBLISHING PTE in Journal of Data Science and Intelligent Systems
- Vol. 3 (2) , 117-125
Comparative Evaluation of Facial Paradigms for Devanagari Script-Based P300 Speller: Preliminary Findings
Display paradigm design is an important stage in the design of any brain–computer interface device including the P300 speller system. Conventional grey-to-white intensification limits the usability of the existing Devanagari script-based P300 speller (DSP3S) as it suffers from reduced visual contrast, diminished user engagement, limited personalization, minimal cognitive stimulation, and the risk of visual fatigue. The use of a facial paradigm can address the above-mentioned limitations through emotionally resonant stimuli, offering diverse cognitive stimulation, and potentially reducing visual fatigue by introducing dynamic and engaging facial stimuli. In this context, this study is the first attempt to investigate the efficacy of facial paradigms in further improving the usability of DSP3S. The authors investigated three facial paradigms based on the smiley face, famous face, and family faces (FFs) for DSP3S. The three facial paradigms are evaluated based on their effectiveness, efficiency, and satisfaction. Effectiveness was measured using classification performance and amplitude difference waveform, whereas efficiency and satisfaction were measured using NASA-TLX, visual analogue scale-fatigue scale, and system usability scale, respectively. The obtained results showed that using the proposed FF intensification, efficiency, and satisfaction could be improved. However, no significant difference was observed in effectiveness among the three display paradigms. In this study, apart from investigating the usability of three display paradigms, authors also propose brain signal decoding using morphological filtering and extreme gradient boosting. Finally, the study concludes that the design of display paradigms using different intensification patterns affects the usability of the speller system and should be taken into consideration. Received: 31 July 2023 | Revised: 20 December 2023 | Accepted: 29 December 2023 Conflicts of Interest Narendra D. Londhe is an Editorial Board Member for Journal of Data Science and Intelligent Systems, and was not involved in the editorial review or the decision to publish this article. The authors declare that they have no conflicts of interest to this work. Data Availability Statement The data that support the findings of this study are not publicly available due to privacy concerns. Data available on request from the corresponding author upon reasonable request. Author Contribution Statement Vibha Bhandari: Conceptualization, Methodology, Software, Investigation, Data curation, Writing - original draft, Visualization. Narendra D. Londhe: Resources, Writing - review & editing, Supervision, Project administration. Ghanahshyam B. Kshirsagar: Validation, Formal analysis, Writing - review & editing.
- 2 January 2024
- Published by BON VIEW PUBLISHING PTE in Journal of Data Science and Intelligent Systems
- Vol. 3 (2) , 117-125
Open AccessComparative Evaluation of Facial Paradigms for Devanagari Script-Based P300 Speller: Preliminary Findings
Display paradigm design is an important stage in the design of any brain–computer interface device including the P300 speller system. Conventional grey-to-white intensification limits the usability of the existing Devanagari script-based P300 speller (DSP3S) as it suffers from reduced visual contrast, diminished user engagement, limited personalization, minimal cognitive stimulation, and the risk of visual fatigue. The use of a facial paradigm can address the above-mentioned limitations through emotionally resonant stimuli, offering diverse cognitive stimulation, and potentially reducing visual fatigue by introducing dynamic and engaging facial stimuli. In this context, this study is the first attempt to investigate the efficacy of facial paradigms in further improving the usability of DSP3S. The authors investigated three facial paradigms based on the smiley face, famous face, and family faces (FFs) for DSP3S. The three facial paradigms are evaluated based on their effectiveness, efficiency, and satisfaction. Effectiveness was measured using classification performance and amplitude difference waveform, whereas efficiency and satisfaction were measured using NASA-TLX, visual analogue scale-fatigue scale, and system usability scale, respectively. The obtained results showed that using the proposed FF intensification, efficiency, and satisfaction could be improved. However, no significant difference was observed in effectiveness among the three display paradigms. In this study, apart from investigating the usability of three display paradigms, authors also propose brain signal decoding using morphological filtering and extreme gradient boosting. Finally, the study concludes that the design of display paradigms using different intensification patterns affects the usability of the speller system and should be taken into consideration. Received: 31 July 2023 | Revised: 20 December 2023 | Accepted: 29 December 2023 Conflicts of Interest Narendra D. Londhe is an Editorial Board Member for Journal of Data Science and Intelligent Systems, and was not involved in the editorial review or the decision to publish this article. The authors declare that they have no conflicts of interest to this work. Data Availability Statement The data that support the findings of this study are not publicly available due to privacy concerns. Data available on request from the corresponding author upon reasonable request. Author Contribution Statement Vibha Bhandari: Conceptualization, Methodology, Software, Investigation, Data curation, Writing - original draft, Visualization. Narendra D. Londhe: Resources, Writing - review & editing, Supervision, Project administration. Ghanahshyam B. Kshirsagar: Validation, Formal analysis, Writing - review & editing.
- 21 June 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cognitive and Developmental Systems
- Vol. 16 (2) , 692-705
TL-P3GAN: An Efficient Temporal-Learning-Based Generative Adversarial Network for Precise P300 Signal Generation for P300 Spellers
Problem of data imbalance among target and non-target classes is inherent in the oddball paradigm-based P300 speller. This class imbalance is a critical issue and requires advanced rigorous learning to deal with. Conventionally, data level-like sampling approaches and algorithmic level-like ensemble approaches had been attempted in past research for data augmentation. However, information loss, overfitting and subject variability were their major pitfalls. Alternatively, generative adversarial network (GAN) based data augmentation managed to alleviate information loss but exhibits problems of overfitting and subject variability due to lack of temporal learning. To compensate for those problems, authors have proposed novel temporal learning-based GAN (TL-P3GAN) to generate precise P300 signals and augment the minority class i.e., P300. The TL-P3GAN comprises novel contribution of multiscale morphological learning in both generator and discriminator. Moreover, the multiscale hybrid model in generator learns multi-resolution morphological information and considers sample-wise latency information from original P300. The effectiveness of TL-P3GAN was confirmed by qualitative and quantitative evaluation metrics with two standard datasets. Further, the work is extended to analyze the effect of generated P300 signals on P300 classification and found significant performance improvement of 8-16% with both datasets in comparison with existing conventional and GAN-based data augmentation approaches.
- 21 June 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cognitive and Developmental Systems
- Vol. 16 (2) , 692-705
TL-P3GAN: An Efficient Temporal-Learning-Based Generative Adversarial Network for Precise P300 Signal Generation for P300 Spellers
Problem of data imbalance among target and non-target classes is inherent in the oddball paradigm-based P300 speller. This class imbalance is a critical issue and requires advanced rigorous learning to deal with. Conventionally, data level-like sampling approaches and algorithmic level-like ensemble approaches had been attempted in past research for data augmentation. However, information loss, overfitting and subject variability were their major pitfalls. Alternatively, generative adversarial network (GAN) based data augmentation managed to alleviate information loss but exhibits problems of overfitting and subject variability due to lack of temporal learning. To compensate for those problems, authors have proposed novel temporal learning-based GAN (TL-P3GAN) to generate precise P300 signals and augment the minority class i.e., P300. The TL-P3GAN comprises novel contribution of multiscale morphological learning in both generator and discriminator. Moreover, the multiscale hybrid model in generator learns multi-resolution morphological information and considers sample-wise latency information from original P300. The effectiveness of TL-P3GAN was confirmed by qualitative and quantitative evaluation metrics with two standard datasets. Further, the work is extended to analyze the effect of generated P300 signals on P300 classification and found significant performance improvement of 8-16% with both datasets in comparison with existing conventional and GAN-based data augmentation approaches.