Research Themes

The CNERLab focuses on research in computational neuroscience and bio-signal analysis, developing advanced hardware solutions for biomedical applications.

1. Brain-Computer Interfaces (BCI)

  • Motor Imagery BCI with Online Adaptation: Development of adaptive BCI systems using motor imagery and error-related potentials (ErrPs) for robust real-time interaction.
  • Feature Extraction & Classification: Techniques include SpecCSP, SVMs, deep networks, and reinforcement learning for error-driven classifier adaptation.
  • Hardware Development: Portable EEG headsets and embedded systems to improve accessibility and usability of BCI applications.

2. Seizure Detection and Prediction

  • CNN/LSTM/Transformer Models: Applied to 2D image representations of EEG data for seizure classification.
  • Rare Event Prediction: Long-term EEG context modeling using attention-based architectures for identifying pre-seizure states.
  • Open Datasets and Tools: Curated large-scale EEG datasets and image generation pipelines for public use.

3. Brain-to-Text Communication

  • Neural Decoding Algorithms: Real-time, low-latency decoding of speech-related neural signals into text using transformer-based models.
  • Generative Language Integration: Error correction using ErrPs and fine-tuned GPT models to enhance sentence fluency.
  • Custom BCI Hardware: Development of EEG-based text generation systems using minimal electrodes and embedded AI modules.