Home

As of July 2024

It might take me a bit to get everything written down, but I'll carve out some time for this section (there's just so much new tech these days!). In the meantime, I've set up a showcase section where you can check out all the source code and demos of the techniques I'll be covering. Stay tuned and happy learning!

Overview

Machine learning is a cornerstone of modern AI, enabling systems to learn from data and make intelligent decisions. By mastering these techniques, you can build models that predict outcomes, uncover patterns, and make autonomous decisions.

Supervised Learning

In supervised learning, you'll dive into algorithms like linear regression, decision trees, and neural networks. For instance, predicting Boston housing prices helps you understand how to train, test, and evaluate models, while finding donors for CharityML introduces more complex algorithms like SVMs and ensemble learning. These skills are essential for tasks where labeled data guides the learning process, from spam detection to medical diagnosis.

Unsupervised Learning and Deep Learning

Unsupervised learning techniques, such as clustering and dimensionality reduction, allow you to find hidden patterns in data without predefined labels. Creating customer segments is a practical example, helping businesses tailor their marketing strategies. Additionally, deep learning, particularly with CNNs, is crucial for tasks like image recognition, as seen in the dog breed classifier project. These methods expand your ability to handle complex data and extract meaningful insights.

Reinforcement Learning

Reinforcement learning teaches you how to train agents to make decisions by interacting with their environment. Training a quadcopter to fly showcases the power of algorithms like Q-learning and policy gradients. This approach is pivotal for developing systems that learn optimal behaviors through trial and error, applicable in robotics, gaming, and autonomous driving.

Overall, these machine learning techniques form the foundation for creating intelligent systems capable of tackling a wide range of real-world problems, making them indispensable in today's data-driven world.

Table of content

• Predicting Boston Housing Prices (Basic Knowledge)
  • Training and testing models
  • Evaluation metrics
  • Valuation and validation
• Find Donors for CharityML (Supervised Learning)
  • Linear regression
  • Perceptron Algorithm
  • Logistic Regression
  • Neural Networks
  • Decision Tree
  • Naive Bayes
  • Support Vector Machines(SVM)
  • Ensemble of learners
• Create Customer Segments (Unsupervised Learning)
  • Clustering
  • Hierarchical & Density-based clustering
  • Gaussian mixture models
  • Feature scaling
  • Dimensionality reduction
• Dog Breed Classifier (Deep learning)
  • Deep Neural Network (DNN)
  • Convolutional Neural Networks (CNN)
• Train a Quadcopter to fly (Reinforcement Learning)
  • Problem - RL framework: Markov Decision Processes
  • Solution - RL framework: Policies, value function and Bellman Equations
  • Dynamic progremming
  • Monte Carlo Methods
  • Temporal - Sarsa, Q-Learning, Expected Sarsa algorithms
  • Work with Continuous spaces
  • Deep Q-learning
  • Policy gradients
  • Actor-critic methods