Computational Neuroscience and Learning Models

MATLAB implementations of foundational models and algorithms in computational neuroscience and machine learning, completed as coursework at Sharif University of Technology.

Topics Covered

Neuron models

• Hodgkin–Huxley model
• Izhikevich spiking neuron model
• Network of leaky integrate-and-fire neurons
• Noisy-output neuron model
• Bifurcation analysis and phase-plane plotting

Spike-train analysis

• Peri-Event Time Histogram (PETH) and raster plots
• Spike-Triggered Average (STA)

Learning and networks

• Hopfield network (auto-associative memory)
• Linear perceptron
• Two-layer neural network for handwritten-digit classification (MNIST-like)
• Reinforcement learning — classical conditioning paradigms
• K-Means clustering (image segmentation, includes a demo on ponyo.jpeg)
• Principal Component Analysis (1D toy data and an eigenfaces demo on faces.mat)

Repository Structure

Bifurcation.m
Hodgkin-Huxley Model.m
Izhikevich Model.m
Linear Perceptron.m
Network of Integrate-and-fire Neurons.m
Noisy output Model.m
Phase Plain.m
Reinforcement Learning - Conditioning Paradigms.m
Hopfield Network/
K-Means/
PCA/
PETH and Raster Plot/
Spike Triggered Average/
two-layer neural network for hand digit problem/

Requirements

• MATLAB (R2019b or later recommended)
• Statistics and Machine Learning Toolbox (for some scripts)
• Image Processing Toolbox (for the K-Means image demo)

How to Run

Open MATLAB in this folder and run any script directly, e.g.:

>> run('Hodgkin-Huxley Model.m')
>> cd('PCA'); run('PCA.m')

Subfolders contain their own .mat data files where needed.