// Publications & Thesis
Research Work
Published at international undergraduate conferences. Research at the intersection
of generative modeling, EEG signal processing, and robust deep learning.
Conference Papers
Publications
01
UCICS 2026 — Undergraduate Conference on Intelligent Computing and Systems
Subject-Independent Event Recognition Using a Lightweight 1D Neural Network
- Designed an efficient 1D neural architecture specifically targeting cross-subject generalization in EEG-based event recognition tasks.
- Achieved 95.35% accuracy under the rigorous leave-one-subject-out (LOSO) cross-validation protocol.
- Addressed the fundamental challenge of subject variability in BCI research through careful architectural choices and regularization.
02
UCICS 2025 — Undergraduate Conference on Intelligent Computing and Systems
Data Augmentation Approach to Frequency Recognition Using Mask-Encoding Based Deep Learning
- Developed a novel mask-encoding based augmentation methodology to improve model invariance under noisy time-series EEG conditions.
- Enhanced classification performance by increasing feature diversity and reducing sensitivity to signal artifacts.
- Demonstrated improved robustness on SSVEP frequency recognition benchmarks, particularly in low-data regimes.
Undergraduate Thesis · Dec 2024 – Oct 2025
GAN-Based Data Augmentation for SSVEP Recognition
Human-Computer Interaction (HCI) Lab · University of Rajshahi · Supervisor: Dr. Md. Ekramul Hamid
- Implemented a conditional WGAN-GP to synthesize realistic EEG time-series data for low-resource BCI tasks.
- Enhanced downstream SSVEP classification performance by increasing synthesized data diversity.
- Optimized GAN training stability through gradient penalties and spectral normalization.
- Applied noise modeling to better match real-world EEG acquisition conditions.
- Conducted extensive multi-user dataset evaluation emphasizing generalization and fairness.
- Reduced subject bias in augmented datasets, improving cross-subject applicability.
Direction
Research Focus
Primary
Generative AI & Large Models
GANs, diffusion models, foundational generative architectures, and their application to scientific data synthesis.
Applied
EEG / BCI Research
Brain-computer interface systems, EEG signal classification, SSVEP recognition, cross-subject generalization.
Theoretical
Representation Learning
Feature disentanglement, learned embeddings, latent space geometry, and robust optimization methods.