Week 13: Introduction to Neural Networks for ML Practitioners

Overview

This week bridges traditional machine learning approaches with neural networks, focusing on parallels between familiar sklearn-style workflows and modern deep learning frameworks like PyTorch and TensorFlow.

Learning Objectives

• Understand neural network fundamentals
• Compare ML and DL workflows
• Implement basic neural networks
• Transition from sklearn to deep learning
• Master basic framework operations

1. Neural Network Fundamentals

• Basic Components
  • Neurons and layers
  • Activation functions
  • Weights and biases
• Comparison with Traditional ML
  • Linear models connection
  • Feature transformation
  • Model complexity
  • Parameter learning

2. Framework Introduction

• TensorFlow
  • Keras interface
  • Model building
  • Training API
  • Data pipelines

3. From Sklearn to Deep Learning

• Workflow Comparison
  • Model instantiation
  • Fitting process
  • Prediction methods
  • Parameter tuning
• Framework Differences
  • Model definition
  • Training approach
  • Data handling
  • Customization options

4. Practical Considerations

• Model Construction
  • Layer selection
  • Architecture design
  • Hyperparameter choice
• Training Process
  • Loss functions
  • Optimizers
  • Batch processing
  • Validation approaches