This repository contains materials for the workshop "Decoding the Brain through Machine Learning", a hands-on introduction to using machine learning techniques for analyzing neural spike trains. Participants will learn how to apply dimensionality reduction, clustering, and classification to identify latent patterns within high-dimensional neural data.
Lucas CS Tavares (lucastavares@neuro.ufrn.br), Rodrigo MM Santiago (rsantiago@neuro.ufrn.br)
This workshop offers a blend of theory and practice in machine learning for neural data analysis. By the end, participants will have a foundational understanding of neural coding principles and practical skills in using Python-based tools to decode brain activity.
-
Introduction to Neural Coding and Signal Types
Understand neural signal scales (macro, meso, micro), key electrophysiological data types (LFPs and spikes), and coding schemes such as rate coding and tuning curves. -
High-Dimensional Neural Data Challenges
Explore the complexities involved in handling large and complex datasets in neuroscience. -
Dimensionality Reduction and Latent Patterns
Learn methods like PCA and UMAP for dimensionality reduction and explore latent patterns with CEBRA. -
Model Evaluation and Validation
Understand metrics like accuracy, ROC-AUC, and F1 score, and perform cross-validation techniques (K-fold, repeated, stratified) for model validation.
- Python (3.7 or higher)
- Jupyter Notebook or Google Colab for running the notebook interactively
neural_ml_decoding.ipynb- The main workshop notebook, containing theory, code exercises, and practical examples.README.md- Workshop overview and instructions.
For questions or feedback, please reach out via the GitHub Issues page, or contact the workshop organizers directly via email.
Happy learning and decoding!