rosoku 0.0.3

A flexible callback-based toolkit for EEG/BCI experiments that lets you define data loading and preprocessing freely, while rosoku handles training, evaluation, and result logging for conventional and deep-learning models.

🕯️ Rosoku — A Flexible EEG/BCI Experiment Pipeline Toolkit

rosoku is a callback-based experiment pipeline for EEG/BCI research.
You are free to design how your data is loaded, processed, and shaped,
while rosoku handles training, evaluation, logging, and result export.

You control the data.
rosoku handles everything after.

Working directly with PyTorch or scikit-learn provides flexibility but requires heavy boilerplate.
Meanwhile, frameworks like MOABB or Braindecode are convenient, but restrict custom processing.

rosoku fills the space between them — flexibility without overhead.

---

🔥 Key Idea

Task	rosoku handles	You define
Dataset loading	receives via callback	how to load (MNE, NumPy, custom)
Preprocessing / feature extraction	pluggable via callbacks	any processing you write
Training loop	model fitting, scheduling, saving	sklearn estimator / PyTorch model
Evaluation & logging	accuracy / saliency	optional W&B configuration
Result export	DataFrame / parquet / msgpack	downstream analysis or plotting

---

🔧 Two Complementary Pipelines

API	Purpose	Typical models
conventional()	traditional ML classification	MDM / TSClassifier / CSP / SVM / LDA
deeplearning()	deep learning with PyTorch	EEGNet / Braindecode / custom CNN/RNN

Both follow the same concept: You write data & preprocessing. Rosoku handles training & evaluation.

---

🚀 Quick Start

Full runnable examples are available under examples/.

Recommended first files:

• examples/example_within-subject-classification-riemannian.py
• examples/example_within-subject-classification-deeplearning.py