This course focuses on transforming your code into a reusable R package and applying it to a real-world problem. You'll refactor your existing components into a structured package, build a Model R6 class for easier network definition and training, and finally, train your neural network on the Boston Housing dataset for a regression task.
Overview
Syllabus
- Unit 1: Building Neural Networks in R
- Debugging Import Errors in a Modular Neural Network Package
- Export Activation Functions and Derivatives in R Package
- Creating the Main Package Initialization File for a Neural Network Library in R
- Unit 2: Modular Training Components
- Define the XOR Dataset for Neural Network Training in R
- Implementing the Training Loop for a Modular Neural Network in R
- Post-Training Evaluation and Results Table for XOR Neural Network
- Unit 3: Model Orchestration in R
- Implementing Model Initialization and Compilation Functions in R
- Defining Abstract Methods and the Predict Interface in an R Model Class
- Implementing the fit Method for Neural Network Training in R
- Implementing the SequentialModel Class in R
- Solving the XOR Problem with a Custom Neural Network in R
- Unit 4: Preparing Real World Data
- Loading and Inspecting the California Housing Dataset in R
- Splitting the Dataset into Training and Testing Sets in R
- Feature Scaling for Neural Network-Ready Data in R
- Unit 5: Real World Neural Network Application
- Fixing Neural Network Output Activation for Regression in R
- Build and Train a Neural Network for House Price Prediction in R
- Train Your Neural Network on California Housing Data
- Evaluate Neural Network Regression Performance on Scaled and Original Data