Machine Learning (ML) Course

This repository is a comprehensive course on machine learning, covering essential topics and their implementations from open classroom and additionally examples. The course is designed to provide both theoretical understanding and practical skills in various machine learning algorithms and techniques. The content follows the Stanford course CS229, taught by Andrew Ng. We augment this course with implementations in Python.

Outlines

---

Regression:

• Linear Regression: modeling relationship between a dependent variable and one independent variable by fitting a linear equation to the observed data.
• Cost Function: evaluating the performance of a ML model by quantifying the difference between the predicted and actual values, guiding the optimization process.
• Optimizations: Techniques used to adjust the parameters of model to minimize the cost function and improve model performance, including gradient descent and other optimization algorithms.
• Mutliple Linear Regression: An extension of linear regression that models the relationship between a dependent variable and multiple independent variables by fitting a linear equation to the observed data.

Classification:

• Logistic Regression: A classification algorithm that models the probability of a binary outcome based on one or more predictor variables using a logistic function.
• Support Vector Classifier: modeling the relationship between input features and continuous target values by finding the optimal hyperplane that minimizes prediction errors.
• Feature Descriptor: Methods for detecting and extracting features from images, such as LBP, SIFT, ORB, HOG and others.
• Feature Based Learning (Annotation): Methods for training various ML algorithm to learn patterns through features.

Artificial Neural Networks (ANN):

Introducing Building Blocks of Artificial Neural Networks:

• Input Layers: The initial layer that receives the input data.
• Hidden Layers: Intermediate layers that process inputs through weighted connections.
• Output Layers: The final layer that produces the network's predictions or outputs.
• Perceptron: A single-layer neural network unit used for binary classification.
• Activation Functions: Functions that introduce non-linearity, allowing the network to learn complex patterns.
• Forward Pass: The process of calculating the output of a neural network by passing input data through each layer of the network.
• Backward Pass (Back Propagation): The method of adjusting the weights of a neural network by propagating the error gradient backward through the network using the chain rule.
• Shallow Networks: Neural networks with few layers, suitable for simpler problems but prone to underfitting.
• Deep Neural Networks: Neural networks with many layers, capable of modeling complex patterns but at risk of overfitting.

Implementation:

Each lecture includes the following implementations:

• Manual Implementation: Concepts are implemented using numpy without relying on built-in machine learning libraries, offering a clear understanding of the underlying mechanics.
• scikit-learn Implementation: Demonstrates basic machine learning models using scikit-learn's built-in functionalities for ease of use and efficiency.
• Keras Implementation: Utilizes the high-level Keras API, which provides user-friendly implementations of specialized machine learning and deep learning models with both sequential and functional approaches.
• PyTorch Implementation: Employs the PyTorch framework, known for its flexibility and research-oriented design, to implement neural networks with full control over their architecture and training process.

Getting Started

1. Clone the Repository:

   git clone https://github.com/qazimsajjad/Machine-Learning-Course.git

2. Navigate to the Directory:

   cd Machine-Learning-Course

3. Install Dependencies:

   python => 3.9
   numpy
   matplotlib
   pillow
   opencv
   sk-learn
   torch
   keras
   tensorflow
   

Usage

Open the Jupyter Notebooks provided in the repository to explore different Machine Learning techniques. Each notebook contains detailed explanations, code implementations, and example images to help you understand the concepts.

Contributor:

Kaleem Ullah Research Assitant Digital Image Processing (DIP) Lab Department of Computer Scinece Islamia College University, Peshawar, Pakistan. Remote Research Assistant Visual Analytics Lab (VIS2KNOW) Department of Applied AI Sungkyunkwan University, Seoul, South Korea.