Skip to content

Folders and files

NameName
Last commit message
Last commit date

Latest commit

ย 

History

12 Commits
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 

Repository files navigation

๐Ÿค– AI Foundations Lab

Python License GitHub stars Last Commit

A professional collection of AI/ML fundamentals, implemented from first principles to advanced applied projects.

Explore Projects โ€ข Key Features โ€ข Connect


โญ Star this repo if you find it useful!

๐Ÿ“Œ Table of Contents


๐Ÿ“– About

This repository serves as a comprehensive laboratory for my journey through the foundations of Artificial Intelligence. Driven by a "Back to Basics" and "First Principles" philosophy, this project documents my transition from academic theory (Stanford/MIT curriculum) to high-fidelity implementation.

Whether it's deriving the gradients for Logistic Regression or architecting a Deep Convolutional GAN, every line of code here is built with rigor, documentation, and a focus on visual performance analytics.


โœ… Key Features

  • โœ… From-scratch implementations: Neural networks and ML algorithms built using only NumPy.
  • โœ… LaTeX math explanations: Detailed derivations for cost functions, loss functions, and optimization algorithms.
  • โœ… Clean documented code: High-quality, professional code with comprehensive comments.
  • โœ… Jupyter visualizations: Real-time plotting of training curves, decision boundaries, and model outputs.
  • โœ… Real-world projects: Applied computer vision (YOLOv8) and generative models (GANs).

๐Ÿš€ Project Showcase

๐Ÿค– Machine Learning

Focuses on classical algorithms, statistical learning theory, and supervised/unsupervised learning fundamentals.

  • Includes: Linear/Logistic Regression, K-Means, PCA, Decision Trees.

๐Ÿง  Deep Learning

Deep dives into multi-layer perceptrons, convolutional neural networks (CNNs), and optimization techniques.

  • Includes: Backpropagation, CNN Architectures, Regularization, Batch Norm.

๐Ÿ’ฌ NLP

exploration of sequences and linguistics in AI.

  • Includes: Word Embeddings (Word2Vec), RNNs, LSTMs, and an introduction to Transformers.

๐ŸŽฎ Reinforcement Learning

Implementing agents that learn from interaction with their environment.

  • Includes: Q-Learning, SARSA, and Deep Q-Networks (DQN).

โญ Applied Projects

This is where the theory meets the real world. This section highlights high-fidelity implementations of state-of-the-art architectures.

Complete inference and training pipeline for YOLOv8, including hardware acceleration (CUDA) and performance analytics.

Implementation of DCGAN for synthetic image generation, exploring adversarial loss and transposed convolutions.


๐Ÿ–ผ๏ธ Showcase Gallery

๐ŸŽฏ Object Detection (YOLOv8)

[!NOTE] > Caption: Real-time object detection with YOLOv8 - detecting people, vehicles, and objects.

YOLO Object Detection Detecting complex scenes with high confidence counts.

YOLO Zidane Human and tie detection on sample imagery.

YOLO Video Frame Real-time video inference frame analysis.

๐ŸŽจ Generative Art (GANs)

[!NOTE] > Caption: Fashion-MNIST: Real vs Synthetic images generated by GAN.

GAN Fashion MNIST Synthetic fashion items generated after adversarial training.


๐Ÿ› ๏ธ Technologies

Python PyTorch TensorFlow NumPy Pandas Scikit-Learn OpenCV Jupyter


๐Ÿ“‚ Repository Structure

ai-foundations-lab/
โ”œโ”€โ”€ 01-machine-learning/          # Classical ML algorithms
โ”œโ”€โ”€ 02-deep-learning/             # Neural Networks & CNNs
โ”œโ”€โ”€ 03-nlp/                       # Natural Language Processing
โ”œโ”€โ”€ 04-reinforcement-learning/    # RL Agents & Environments
โ”œโ”€โ”€ 05-applied-projects/          # SOTA Architectures (YOLO, GANs)
โ”‚   โ”œโ”€โ”€ Generative Adversarial Networks/
โ”‚   โ””โ”€โ”€ Object-Detection-YOLO/
โ”œโ”€โ”€ assets/                       # Repository images & media
โ””โ”€โ”€ requirements.txt              # Project dependencies

๐Ÿ—บ๏ธ Roadmap / Future Work

  • Vision Transformers (ViT): Implementation and comparison with CNNs.
  • More RL environments: Training agents on Mujoco/Gymnasium environments.
  • LLM fine-tuning: Experiments with LoRA and QLoRA on open-source LLMs.
  • Web app deployments: Serving models via FastAPI and Docker.

๐Ÿ Getting Started

  1. Clone the repository:

    git clone https://github.com/DaviBonetto/ai-foundations-lab.git
    cd ai-foundations-lab
  2. Set up the environment:

    python -m venv .venv
    source .venv/bin/activate  # On Windows: .venv\Scripts\activate
    pip install -r requirements.txt
  3. Explore the notebooks: Launch Jupyter and dive into any module!


๐Ÿ“š Learning Resources

This project is inspired by and follows the academic rigor of graduate-level computer science courses:

Stanford University:

  • CS229: Machine Learning - Supervised learning, unsupervised learning, learning theory
  • CS230: Deep Learning - Neural networks, CNNs, RNNs, optimization techniques
  • CS224N: Natural Language Processing with Deep Learning - Word embeddings, transformers, attention mechanisms
  • CS234: Reinforcement Learning - MDPs, Q-learning, policy gradients, deep RL
  • CS221: Artificial Intelligence: Principles and Techniques - Search, reasoning, learning
  • CS336: Language Models - Transformer architecture, LLM foundations

Additional Resources:

  • MIT 6.S191: Introduction to Deep Learning
  • Research papers and textbooks by leading AI researchers

Note: This is a self-directed learning project using publicly available course materials. No formal enrollment or academic credit.


๐Ÿค Connect With Me

LinkedIn GitHub Email

"The only way to learn a new programming language is by writing programs in it." โ€” Dennis Ritchie


Developed by Davi Bonetto

About

A comprehensive self-directed learning journey through Machine Learning, Deep Learning, NLP, and Reinforcement Learning - built from first principles

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages