Human-memory

A repository for computational models of human memory

Projects are implemented in Javascript and Python.

1) Self organizing map/ Kohonen net (implemented in Javascript)

About SOMS: A self organizing map works by creating a field of several different classification populations through competive training.

Run instructions: Open index.html

2) Hopfield net

About Hopfield Nets: Finds the local minimums in a global energy function through activity-correlation. When fully trained, this allows for content-adressable memory.

Run Instructions: Move to server directory, and run the command python server.py.
Next, open http://127.0.0.1:5000/

3) Restricted Boltzman Machines

Includes the RBM: Most of the credit for this code goes to Edwin Chen, I simply extended it to work in a 2D visual environment.

[Edwin Chen Tutorial link] (https://github.com/echen/restricted-boltzmann-machines)

4) Models to implement

• Spiking neuron models
• Sparse encoding (see Numenta)
• Reinforcement learning models
• Neural networks with backprop
• Convolutional neural networks