DEFT: Disentanglement-Enhanced Fine-Tuning for EEG Foundation Models
EEG foundation models (EFMs) have improved EEG representation learning by transferring knowledge from large-scale multi-subject pretraining to downstream tasks.
ProofPaper ↗
Key points
- To address this issue, we propose Disentangled EFM Fine-Tuning (DEFT), a plug-and-play adapter for EFM fine-tuning.
- DEFT encourages branch disentanglement of pretrained tokens into task-oriented content and subject-consistent style, then performs downstream prediction from the content branch.
- It instantiates this factorize-before-classify design with recursive Gaussian refinement, temporal-variance regularization, subject-aware contrastive learning, and reconstruction.
- Experiments across multiple EEG datasets and four EFM backbones show that DEFT improves matched fine-tuning baselines.
Sources (1)
- [1]DEFT: Disentanglement-Enhanced Fine-Tuning for EEG Foundation ModelsarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 10, 02:54 AM
EEG foundation models (EFMs) have improved EEG representation learning by transferring knowledge from large-scale multi-subject pretraining to downstream tasks.
To address this issue, we propose Disentangled EFM Fine-Tuning (DEFT), a plug-and-play adapter for EFM fine-tuning.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 10, 2026Co-adaptive Rule Routing and Constraining for Structured Knowledge Retrieval
- Oct 9, 2026AI Models Can Provably Hide Arbitrary Capabilities
- Oct 9, 2026Minions: A Dynamic and Distributed Virtual GPU Cloud Built on Idle Apple Silicon Devices
- Oct 9, 2026Scaling Agentic Data for Long-Horizon Terminal Intelligence
- Oct 8, 2026Understanding Jev, the new model everyone is talking about
- Oct 8, 2026Beyond Imitation: A Framework and Benchmark for LLM-Assisted Peer Review