Foundation of Machine Learning & Neural Architectures (MLOM Lab Activities)

This repository is a comprehensive showcase of machine learning algorithms, data engineering workflows, and deep learning implementations. It transitions from fundamental Pythonic data structures to complex neural network architectures using TensorFlow and Keras.

🚀 Key Learning Modules

1. Neural Networks & Deep Learning (TensorFlow/Keras)

• MLP Classification: Building and tuning Multi-Layer Perceptrons to predict health outcomes (Diabetes dataset).
• Deep Regression: Implementing neural networks for price prediction using the Boston Housing dataset, focusing on Root Mean Squared Error (RMSE) optimization.
• Architecture Design: Experimenting with layer depth, dropout, and activation functions for optimized convergence.

2. Algorithmic Implementations (From Scratch & Scikit-Learn)

• Support Vector Machines (SVM): A mathematical deep-dive into SVMs. Includes custom optimization using scipy.minimize and visualization of decision boundaries on the Iris dataset.
• Linear Regression: Statistical modeling of relationships (e.g., Salary vs. Experience) using both statsmodels and sklearn.

3. Data Engineering & Preprocessing

• Scaling & Standardization: Detailed comparisons between StandardScaler, MinMaxScaler, and Robust scaling techniques to handle variance and outliers.
• Advanced Pythonics: High-performance data manipulation using List Comprehensions and optimized NumPy operations.
• Outlier Management: Systematic detection and replacement strategies to ensure model stability.

4. Specialized Domains

• Computer Vision: Basic image processing and edge detection using OpenCV (cv2), specifically implementing Canny Edge Detection.
• Information Retrieval: Implementation of an Inverted Index—the foundational logic behind search engines—mapping terms to document IDs.

🛠 Tech Stack

• Core: Python 3.x, NumPy, Pandas
• Modeling: Scikit-Learn, TensorFlow, Keras, Statsmodels
• Visualization: Matplotlib, Seaborn
• Utilities: OpenCV, Scipy (Optimization)

📊 Highlights of Work

Component	Technique	Dataset
Classification	MLP / SVM	Iris, Diabetes
Regression	Neural Networks / OLS	Boston Housing, Salary Data
Computer Vision	Canny Edge Detection	Image Buffers
Indexing	Inverted Indexing	Text Corpora

🧪 Laboratory Breakdown

• MLOM_Lab02 (A/B): Data cleaning, handling missing values, and OpenCV basics.
• MLOM_Lab03/04: Deep dive into Neural Network classification and regression with Keras.
• SVM Lab: Mathematical implementation of maximal margin classifiers.
• Indexing Guide: Logic for term mapping and text processing.

🚀 How to Navigate

Each folder/notebook is self-contained. To replicate the results:

1. Ensure tensorflow and scikit-learn are installed.
2. Follow the sequence from Lab_001 (Fundamentals) to MLOM_Lab04 (Deep Learning).