Skip to content

README.md

Latest commit

 

History

History
92 lines (61 loc) · 3.27 KB

README.md

File metadata and controls

92 lines (61 loc) · 3.27 KB
Raw

🧠 Machine Learning & Digital Image Processing – Academic Projects

This repository includes my academic work in Machine Learning and Digital Image Processing. It consists of practical implementations, conceptual notebooks, and lab-style explorations developed as part of my coursework.

📁 Contents

🔍 Machine Learning Projects

1. Depression_detection_with_neural_nets.ipynb

A neural network–based approach to detecting depressive sentiment in text data. Focuses on preprocessing, feature extraction, and model training using deep learning.

2. EDA_using_python.ipynb

Exploratory Data Analysis (EDA) notebook using pandas, matplotlib, and seaborn. Analyzes data distributions, outliers, and trends.

3. Text_to_SQL.ipynb

A basic implementation for converting natural language questions into SQL queries. Includes tokenization, classification, and SQL query generation.

🔁 Cross-Validation Techniques

This section provides a concise overview of key cross-validation techniques used in model evaluation:

  1. K-Fold Cross-Validation

    • Splits the dataset into k equal-sized folds.
    • Each fold is used once as the validation set, while the remaining k-1 folds form the training set.

  2. Hold-Out Cross-Validation

    • Splits the dataset into separate training and testing sets (e.g., 80/20 or 70/30).
    • Simple and fast but less reliable due to single random split.

  3. Stratified K-Fold Cross-Validation

    • Extension of K-Fold that maintains class distribution across all folds.
    • Especially useful for imbalanced datasets.

  4. Leave-P-Out Cross-Validation (LPOCV)

    • Repeatedly trains the model using all data except p samples, which are used for testing.
    • Computationally intensive but exhaustive.

  5. Leave-One-Out Cross-Validation (LOOCV)

    • A special case of LPOCV where p = 1.
    • Each sample is used once as a test set; best for small datasets.

  6. Monte Carlo (Shuffle-Split) Cross-Validation

    • Randomly splits data into training and testing sets multiple times.
    • Flexible but may lead to test sets overlapping.

  7. Time Series (Rolling) Cross-Validation

    • Used when data is time-dependent.
    • Training set includes past data, and the test set includes future data to preserve temporal order.

📘 Notebook : cross_validation_techniques.ipynb

🖼️ Digital Image Processing (DIP) Projects

This section includes foundational techniques used in image preprocessing and analysis.

Topics Covered:

  • Image Enhancement: Histogram Equalization, Contrast Stretching
  • Noise Removal: Gaussian Blur, Median Filtering
  • Edge Detection: Sobel, Prewitt, Canny Operators
  • Morphological Operations: Dilation, Erosion, Opening, Closing
  • Color Space Conversions: RGB ↔ Grayscale, HSV
  • Frequency Domain Analysis: Fourier Transform
  • Image Segmentation: Thresholding, Region Growing, Contour Detection

📘 Notebook : dip_lab_experiments.ipynb

🛠️ Tech Stack

  • Python
  • Jupyter Notebook
  • pandas, NumPy, matplotlib, seaborn
  • scikit-learn, TensorFlow / PyTorch
  • OpenCV (for image processing)

🚀 Getting Started

  1. Clone the repository:
git clone https://github.com/ReVuz/Machine_Learning.git
cd Machine_Learning