Latent-Preserving MEG Sensor Selection
Developing representation-aware sensor selection methods that preserve teacher latent spaces rather than relying solely on attribution rankings.
About
I study neural representation and decoding using ECoG, MEG, and machine learning approaches.
My research focuses on neural decoding, speech and music BCI, latent representation learning, cross-session adaptation, and biologically meaningful modeling of neural dynamics.
I am particularly interested in how deep neural networks can reveal distributed neural representations beyond conventional feature engineering and connectivity measures.
Research Interests
- Neural decoding and representation learning
- Brain-computer interfaces (BCI)
- Speech and music neurotechnology
- ECoG and MEG signal analysis
- Cross-session and cross-subject adaptation
- Latent geometry of neural dynamics
Selected Projects
Speech & Music Brain-Computer Interfaces
Investigating decoding and reconstruction of imagined speech and music using ECoG recordings.
Cross-Session Neural Adaptation
Modeling neural drift and session variability using covariance geometry, latent alignment, and adaptive decoding.
Publications
Publication list coming soon.
Contact
Email: jii.kwon125@gmail.com