Neural Network Implementation in Go

This repository contains a Go (Golang) implementation of a neural network, inspired by the concepts in the book Neural Networks from Scratch in Python by Sentdex. The goal is to translate these ideas into Go, focusing on a simple yet functional neural network designed to identify patterns in the CAN dataset.

Overview

This project implements a neural network capable of classifying data from the CAN dataset. It covers essential neural network topics such as:

• Building a feedforward neural network from scratch
• Implementing backpropagation and gradient descent for learning
• Training and testing the network using the CAN dataset
• Handling activation functions like ReLU and Softmax
• Evaluating model performance with metrics like accuracy, loss

Features

• Simple Feedforward Neural Network: A basic neural network architecture, configurable with various layers and neurons.
• Activation Functions: Implementation of essential activation functions like ReLU, Sigmoid, and Softmax.
• Backpropagation and Gradient Descent: Algorithms to optimize the network’s weights based on loss.
• Training on CAN Dataset: Code to train the neural network on the IDS dataset for intrusion detection tasks.
• Evaluation and Metrics: Tools to assess the model's performance, focusing on accuracy and loss.

Getting Started

Prerequisites

• Go: Make sure Go is installed on your system. Download it from the official website.

Installation

Clone this repository:

git clone https://github.com/saent-x/IDS-NN.git
cd IDS-NN

Install dependencies:

go mod tidy

Running the Project

To run the neural network with the CAN dataset:

go run main.go