machine-learning-nutsnbolts

The nuts and bolts of machine learning and insights into its operations. Generated by DALL·E 3.

This repository aims to organize implementations of machine learning algorithms and demonstrate practical ways to apply or use machine learning models. I initiated this project to refresh and reinforce my fading knowledge of previously learned concepts, deepen my understanding of various algorithms, and examine examples of their practical applications.

Algorithms

Here is an overview of machine learning algorithms that are either currently implemented or planned for future implementation. Additionally, this table presents supplementary materials I have prepared to revisit and strengthen my understanding of what I learned.

Notes: The "Colab" links will take you to the Google Colab notebook, where you can view the implementation and run the code. The "DEV" column contains links to articles I have written.

Category	Algorithms	Colab	DEV
Optimization	Exploring Gradient Descent Variants: Batch, Mini-Batch, Stochastic, and Fundamentals of Implementation
Optimization	Comparing Optimizers: Gradient Descent, Momentum, AdaGrad, and Adam Implemented from Scratch
Preprocessing	Implementing Feature Scaling Methods: Standardization (Z-score normalization) and Normalization (Min-Max scaling)
Supervised	Implementing Linear Regression from Scratch: Gradient Descent, Mean Squared Error (MSE) Loss, and L1/L2 Regularization Explained
Supervised	Implementing Logistic Regression for Classification from Scratch: Cross-entropy loss, One-hot encoding, and Activation function (Sigmoid/Softmax) Explained
Supervised	Implementing Neural Network for Classification with ReLU and Backpropagation from Scratch to PyTorch
Supervised	Implementing K-Nearest Neighbor (k-NN) for Classification from Scratch
Unsupervised	Implementing K-means for Clustering from Scratch
Unsupervised	Implementing Principal Component Analysis (PCA) for Dimensionality Reduction from Scratch
Reinforcement	Implementing Multi-Armed Bandit (MAB) from Scratch

The above implementations are based on the following resources:

• https://www.coursera.org/learn/machine-learning
• https://github.com/oreilly-japan/deep-learning-from-scratch
• https://www.oreilly.com/library/view/introduction-to-machine/9781449369880

Examples

Below are practical examples demonstrating the application of machine learning algorithms to address real-world challenges, as well as insights into the operation of machine learning models.

Example	Implementations	Supplementary Links
NVIDIA Triton Inference Server

Prerequisites

To experiment with the examples, you will need to follow these installation and setup procedures:

Library	Setup Instructions
Pyenv	https://github.com/pyenv/pyenv
Poetry	https://python-poetry.org/docs

Configuration of the Python version

Ensure that this version aligns with the version used in Poetry, then choose the appropriate version:

$ pyenv install 3.9
$ pyenv local 3.9

Install the required dependencies

The minor version depends on the version of the Python you have installed:

$ poetry env use ~/.pyenv/versions/3.9.17/bin/python3.9
$ poetry install