Introduction to neural networks

Introduction

To learn how machine learning and neural networks work, I’ve decided to build a neural network almost from scratch myself that analyzes realtime images of guitar fretboards.

This guide follows sentdex’s course on youtube, so if you prefer a video tutorial, take a look at his playlist.

Requirements

1. Python - although not my favourite language, Python’s packages and overall simplicity makes it a very popular choice for neural networks and machine learning. If you want to use another language, please do some research yourself and find appropriate tools.
2. Numpy(2.2.3) - a widely used package for data science in Python.
3. Matplotlib(3.10.1) - a package used for visualizing datasets and graphs in Python.
4. OpenCV - a package for ‘vision’.
5. A good understanding of math, especially matrices, linear algebra and calculus. Although I’ll try my best to explain, your prior knowledge of these topics will be extremely beneficial.

Contents

1. Neural networks & neurons
2. The dot product
3. Hidden layers
4. Activation functions
5. Softmax activation