Planar Data Classification with One Hidden Layer

The assignment involves building a shallow neural network with one hidden layer to classify non-linearly separable 2D data using NumPy from scratch.
This repository contains my implementation of the Week 3 programming assignment from the Deep Learning Specialization (Course 1: Neural Networks and Deep Learning) by Andrew Ng on Coursera.

Description

In this assignment, I implemented a neural network architecture from scratch, including both forward and backward propagation, to classify a planar dataset. This model shows how adding non-linearity through activation functions in hidden layers allows for solving problems logistic regression cannot.

Key Concepts Covered:

• Shallow neural networks with one hidden layer
• tanh and sigmoid activation functions
• Binary classification vs non-linear boundaries
• Vectorized implementation of forward and backward propagation
• Cost function and parameter optimization with gradient descent
• Decision boundary visualization

Technologies Used

• Python 3
• NumPy
• Matplotlib (for visualization)

Files

• planar_data_classification.ipynb: Jupyter notebook with the complete assignment
• testCases.py: Helper script for testing your functions
• planar_utils.py: Utility functions for data handling and visualization
• datasets/: Contains the 2D classification dataset

⚠️ This repository includes only my implementation and does not contain auto-graded files. It follows Coursera’s Honor Code.

Course

Part of:

Deep Learning Specialization
Instructor: Andrew Ng

License

This repository is for educational and portfolio purposes only. Please do not use it for submitting assignments directly.

---

If you're learning deep learning too, consider starring this repository!