NeuroDive - A Machine Learning system to decode a monkey’s arm movements from its neural activity

This code is a group coursework submitted as part of the Brain-Machine Interfaces module at Imperial College London.

---

🧐 How does it work?

The proposed neural decoder uses a 2-stage approach:

1. A classifier predicts the reaching angle for the current trajectory.
   Linear Discriminant Analysis (LDA) is used for classification.
2. A linear regressor (trained on data from the relevant reaching angle) predicts position.
   Principal Component Analysis (PCA) followed by Linear Regression (LR) is used for hand position estimation – a technique called Principal Component Regression (PCR)

---

Task setup

Each colour represents a different reaching angle.

---

Visualizing the results

---

💻 Usage

• positionEstimatorTraining.m: trains the model
• positionEstimator.m : predicts one hand position using the trained model
• testFunction_for_students_MTb.m: evaluates model performance on unseen test data