Creating a Neural Network Architecture with GoLang

This was an assignment for the course CSE-465 at North South University. The task was to create a neural network architecture using Go-Lang.

Task

• Determine a representation of a computational graph
• Implement a method to mark the trainable paramters
• Implement an initialization of said parameters
• Implement a method to provide the training data
• Implement backpropagation and gradient descent
• Assume MSE loss
• Must be an implementation for both vector and scalar values
• Implement a CNN filter of any dimension
• Implement Max Pooling

How to run the code:

For the scalar version:

go run scalar_nn.go

For the vector version:

cd v2
go run vector_nn.go