Khin Pyae Phyo San
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Khin Pyae Phyo San has a total of 4 co-authors. Here are the top co-authors by number of shared publications:
- Tetsuya Shimamura
Saitama University
2 shared publications Nwe Nwe Myint Thein
University of Information Technology,Faculty of Computer Science,Yangon,Myanmar
1 shared publication- Nwe Nwe Myint Thein
University of Information Technology
1 shared publication - Yosuke Sugiura
Saitama University
1 shared publication
Latest Publications
- Open Access
- 12 November 2026
- Published by MDPI AG in Preprints.org
Cross-Condition Transfer Learning for EEG-Based Inner Speech Classification: A Controlled Evaluation of Neural Networks and Temporal Segmentation
Transferring a decoder trained on overt (pronounced) speech to inner speech has been proposed to offset scarce inner-speech EEG. This was re-examined on the “Thinking out loud” dataset (ds003626, 10 participants, four words), with no test trial in training or selection. EEGNet, ShallowConvNet, CSP+LDA and a Riemannian tangent-space pipeline were compared on bidirectional transfer, combined training, overt pre-training with fine-tuning on 25–100% inner-speech subsets, maximum-mean-discrepancy alignment and four trial windows. Shuffled-label controls and an exploratory, post hoc test family were used. Cross-participant, held-out-session, channel, band, external-electrode and second-dataset (ds006104) arms were reference checks. Within participant, overt speech was decodable in the action interval (0.34–0.43, chance 0.250). It survived a held-out session, peaked in 30–100 Hz and did not cross participants. Two lip electrodes alone decoded the pronounced word at 0.67–0.83. Regressing the external electrodes out of the scalp reduced three decoders’ overt accuracy to 0.28–0.40, leaving a residual of undetermined origin. Inner-speech accuracy was not distinguishable from chance. No transfer method, fine-tuned or not, improved reliably on its controls. Effects below about 2–6 percentage points were likely missed. These findings suggest that articulation-related features learned from overt speech may not provide sufficiently transferable information for inner-speech decoding under the present conditions.
- 12 November 2026
- Published by MDPI AG in Preprints.org
Open AccessCross-Condition Transfer Learning for EEG-Based Inner Speech Classification: A Controlled Evaluation of Neural Networks and Temporal Segmentation
Transferring a decoder trained on overt (pronounced) speech to inner speech has been proposed to offset scarce inner-speech EEG. This was re-examined on the “Thinking out loud” dataset (ds003626, 10 participants, four words), with no test trial in training or selection. EEGNet, ShallowConvNet, CSP+LDA and a Riemannian tangent-space pipeline were compared on bidirectional transfer, combined training, overt pre-training with fine-tuning on 25–100% inner-speech subsets, maximum-mean-discrepancy alignment and four trial windows. Shuffled-label controls and an exploratory, post hoc test family were used. Cross-participant, held-out-session, channel, band, external-electrode and second-dataset (ds006104) arms were reference checks. Within participant, overt speech was decodable in the action interval (0.34–0.43, chance 0.250). It survived a held-out session, peaked in 30–100 Hz and did not cross participants. Two lip electrodes alone decoded the pronounced word at 0.67–0.83. Regressing the external electrodes out of the scalp reduced three decoders’ overt accuracy to 0.28–0.40, leaving a residual of undetermined origin. Inner-speech accuracy was not distinguishable from chance. No transfer method, fine-tuned or not, improved reliably on its controls. Effects below about 2–6 percentage points were likely missed. These findings suggest that articulation-related features learned from overt speech may not provide sufficiently transferable information for inner-speech decoding under the present conditions.
- 2 August 2026
- Published by Springer Nature
- p. 505-517
Comparative Analysis of Neural Network Architectures for EEG Speech Classification
- 2 August 2026
- Published by Springer Nature
- p. 505-517
Comparative Analysis of Neural Network Architectures for EEG Speech Classification
- 8 February 2025
- Published by Springer Nature
- p. 516-525
Cloud Cost Prediction with LSTM on Parametric Data
- 8 February 2025
- Published by Springer Nature
- p. 516-525
Cloud Cost Prediction with LSTM on Parametric Data
- 7 November 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE)
DLOps: Developing Deep Learning Aided Burmese Rice Price Forecasting
—In nowadays technology areas, a pivotal concern has emerged. It’s was especially concerning architecting Deep Learning enabled Systems (DLS) and sustainability. Most of DLS were failed while release to production. That’s because of uncertainties in data variations, evolutionary requirements, and model inconsistencies. The one suitable solution is Deep Learning Operations (DLOps). It enhances reproducibility and sustainability in building DLS. It still has challenges and are related to environmental impact, technical maintenance, and economic concerns. The self-adaptation helps to address and overcome these challenges. In this paper proposed a sustainable modularized DLOps pipeline thorough the conceptual MAPE-K Framework to solve technical maintenance and economic concerns. The proposed system addresses the uncertainties in data handling and model training then try to proof out efficient utilization from economic concerns point of view. Further, the proposed system is implemented for a Burmese Rice Price Prediction use case to display DLOps pipeline sustainability from technical maintenance point of view.
- 7 November 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE)
DLOps: Developing Deep Learning Aided Burmese Rice Price Forecasting
—In nowadays technology areas, a pivotal concern has emerged. It’s was especially concerning architecting Deep Learning enabled Systems (DLS) and sustainability. Most of DLS were failed while release to production. That’s because of uncertainties in data variations, evolutionary requirements, and model inconsistencies. The one suitable solution is Deep Learning Operations (DLOps). It enhances reproducibility and sustainability in building DLS. It still has challenges and are related to environmental impact, technical maintenance, and economic concerns. The self-adaptation helps to address and overcome these challenges. In this paper proposed a sustainable modularized DLOps pipeline thorough the conceptual MAPE-K Framework to solve technical maintenance and economic concerns. The proposed system addresses the uncertainties in data handling and model training then try to proof out efficient utilization from economic concerns point of view. Further, the proposed system is implemented for a Burmese Rice Price Prediction use case to display DLOps pipeline sustainability from technical maintenance point of view.