The MLP Architecture: Activations & Initialization in R

Overview

This course builds upon single layers to construct a complete Multi-Layer Perceptron (MLP). You'll learn to stack layers, explore different activation functions like ReLU and Softmax, and understand the importance of weight initialization for effective training.

Syllabus

• Unit 1: The MLP Architecture: Activations & Initialization
Implementing Forward Propagation in a Multi-Layer Perceptron
Fixing Layer Dimensions in a Multi-Layer Perceptron
Building a Multi-Layer Perceptron Function in R
Expanding an MLP with an Additional Layer
Building a Multi-Layer Perceptron from Scratch in R
• Unit 2: ReLU Activation and Flexible Layer Design in R MLPs
Fixing the ReLU Activation Function for Matrix Inputs
Implementing ReLU Activation in Your Neural Network
Implementing the ReLU Activation Function in R
• Unit 3: Output Layer Activation Functions: Softmax and Linear in R MLPs
Implementing Numerically Stable Softmax in R
Verifying Softmax Outputs as Valid Probability Distributions
Implementing the Linear Activation Function for Regression Tasks
Debugging Output Activation Functions in Neural Networks
Building Neural Networks with Classification and Regression Output Activations
• Unit 4: Weight Initialization Strategies for Neural Networks in R
Implementing Random Scaled Weight Initialization for Neural Networks
Fixing He Uniform Weight Initialization for Neural Networks
Implementing Xavier Normal Weight Initialization for Neural Networks
Implementing He Uniform Weight Initialization in Neural Networks

---

Subjects

Artificial Intelligence